View
1
Download
0
Category
Preview:
Citation preview
A DEMOGRAPHIC ANALYSIS OF LESSER PRAIRIE-CHICKEN POPULATIONS IN SOUTHWESTERN KANSAS: SURVIVAL, POPULATION VIABILITY, AND
HABITAT USE
by
CHRISTIAN ANDREW HAGEN
B. A., Fort Lewis College, 1993
M.N.R.M., University of Manitoba, 1999
----------------------------------------------
A DISSERTATION
submitted in partial fulfillment of the
requirements for the degree
DOCTOR OF PHILOSOPHY
Division of Biology College of Arts and Sciences
KANSAS STATE UNIVERSITY
2003
Approved by:
Major Professor Robert J. Robel
ABSTRACT
Lesser prairie-chicken (Tympanuchus pallidicinctus) habitat and populations have
been reduced range-wide by more than 90% since the turn of 20th Century. Population
indices in Kansas reflected the range-wide trends. The rate of habitat loss slowed
considerably starting in the 1980s, but populations have continued to decline in the state.
To aid in the conservation of this “warranted but precluded” threatened species, more
information is needed on the basic and applied population ecology of this prairie grouse.
The present research was initiated to collect field data for 3-years and synthesize 6-years
of data from 2 Federal Aid projects in southwestern Kansas.
I used age-structured mark-recapture models to estimate the local survival rates of
banded yearling and adult male lesser prairie-chickens from live mark-recapture data.
Local survival rates of male lesser prairie-chickens were ranked: yearling (φ1 = 0.615, SE
= 0.068) > adult (φ1 = 0.485, SE = 0.058) > older adults (φ2 = 0.347, SE = 0.047).
Using joint models of live encounter and dead recovery, I examined the potential
for bias in survival estimates of radiomarked male lesser prairie-chickens. The model best
supported by the data, Ŝc, pgroup+t, rg, Fc, indicated that survival was best modeled as not
different (Ŝc = 0.731, SE = 0.072) across radiomarked and banded birds.
I evaluated the effects of season, age, and gender on survival of radiomarked
birds. The known-fate analysis revealed that overall male (Ŝ = 0.71, SE = 0.06) and hen
(Ŝ = 0.69, SE = 0.06) survival rates were similar, but hens were most susceptible to
mortality during nesting. Additionally, yearling females had a greater probability (Ŝ =
0.77, SE = 0.06) of surviving than adults (Ŝ = 0.62, SE = 0.05).
Population viability and management alternatives were examined using elasticity
analysis on an age-specific projection matrix. The model was parameterized with
demographic data from this field study. The rate of population change (λ) was < 1.0 for
both populations (λI = 0.544, λII = 0.754). This indicated a short-term decline in
population growth in the absence of immigration. However, the marked contrast in the
contributions to λ between populations were explained by nest success and chick
survival, and prescribed management practices should focus on these rates.
I examined the relationship of several habitat characteristics and landscape
features as they pertained to habitat suitability in southwestern Kansas. I quantified these
characteristics in use and non-use sites as determined from the presence or absence of
prairie chicken locations. Multivariate analyses indicated that site occupancy was
explained, in part, by moderate densities of sagebrush, but negatively associated with the
proximity to anthropogenic features (e.g., powerlines, roads, buildlings, and pump-jacks).
TABLE OF CONTENTS List of figures………….…………………………….…………..…………………..……iv
List of tables………………………………………..……………………………..…..…vii Acknowledgements……………………………………………….…………………..…...x Introduction………………………………………………………….…………………….1 Chapter 1: Age-specific survival of male lesser prairie-chickens…...…………………..16
Introduction…………………….....………………………………………….......17
Study Area……………………….…………………………………………........20
Field Methods……….......………..…….…………………………………….….20
Data Analysis………………………………………………..…………………...21
Results…………………………..………………………………………………..24
Discussion………………………………………..………………………………25
Literature Cited…………………………………………..……………….………31
Chapter 2: Radiotelemetry estimates of survival in the lesser prairie-chicken: are there
biases?……………………………………………………………………………45
Introduction…………………….....………………………………………….......46
Field Methods……….......………..…….…………………………………….….48
Data Analysis………………………………………………..…………………...49
Results…………………………..………………………………………………..52
Discussion………………………………………..………………………………54
Literature Cited…………………………………………..……………….………57
i
Chapter 3: Gender and age-specific survival and probable causes of mortality in
radiomarked lesser prairie-chickens in Kansas…...……………………………..65
Introduction…………………….....………………………………………….......66
Study Area……………………….…………………………………………........68
Field Methods……….......………..…….…………………………………….….69
Data Analysis………………………………………………..…………………...71
Results…………………………..………………………………………………..73
Discussion………………………………………..………………………………76
Literature Cited…………………………………………..……………….………83 Chapter 4: Lesser prairie-chicken demography: a sensitivity analysis of population
dynamics in two prairie fragments…...………………………………………...104
Introduction…………………….....………………………………………….....105
Study Area……………………….…………………………………………......108
Field Methods………......………..…….…………………………………….…109
Data Analysis………………………………………………..……………..…..110
Results…………………………..…………………………………….………..116
Discussion………………………………………..………………..…….…..…120
Literature Cited…………………………………………..……………….……128
Chapter 5: The effects of landscape features on lesser prairie-chicken habitat use……145 Introduction…………………….....………………………………………….....146
Study Area……………………….…………………………………………......147
Field Methods……….......………..…….…………………………..……….….148
Data Analysis………………………………………………..………………….149
ii
Results………………………..………………………………..……………….154
Discussion……………………………………………..…..……………………156
Literature Cited……..…………………………………..……………….………159 Overall Summary………………………………………….……………………..…….173
Appendix I: Genetic evaluation of lesser prairie-chickens….………………………….179
Appendix II: Tables of lesser prairie-chicken morphometrics.…………..…………….198
iii
LIST OF FIGURES Introduction Fig. 1. Historic and current distribution of lesser prairie-chickens in Kansas. Note extra-
limit records for several studies denoted by small symbols (see map legend for details)………………………………………………………………………………13
Fig. 2. Population trends (birds/mi2) of lesser prairie-chickens from lek survey data
(1964-2002) from 4 to10 survey routes (A). The relationship between lek survey data and available rangeland for counties where surveys were conducted (B). Five-year intervals were used because US Census of Agriculture quantified available pasture at those approximate intervals. Lek survey data were averaged across each 5-year interval to illustrate the potential long term effects of large-scale habitat loss………………………………………………………………………………….15
Chapter 1
No figures Chapter 2 Fig. 1. The cumulative frequency distribution (A) of mortalities and right-censoring
indicates that the high mortality periods (2-week intervals) were not associated with right censoring. The distribution of mortality and right-censoring (B) indicates the actual numbers lost to mortality or signal in a given 2-week period were not usually related……………………………………………………………………………….64
Chapter 3 Fig. 1. Annual variation in monthly survival rates (9-months) of hens (A) from the model
S year + month, standard errors not included for clarity. Yearling and adult hen survival rates (B) from the model S year + month. Male and hen survival (C) from the model S gender*month (1997-1999)…………………………………………………………...95
Fig. 2. Relationship between biweekly survival rates of males (dashed line) and hens
(solid line) and the cumulative frequency distribution (CFD) of incubating hens (gray). Note the inverse relationship of hen survival relative to the CFD of nests...97
Fig. 3. Monthly survival rates of males and hens (A) for 12-months (2000-2003) from the
model S gender*month. Note the variation in the timing of survival between the groups. Monthly survival rates for yearlings and adults (B) for 12-months.………….…….99
Fig. 4. Nest success (black) and hen survival (white) ± SEs for 1998-2002. Nests from
1997 were not included due to different nest marking techniques..…………….101
iv
Chapter 3 (continued) Fig. 5. Reproductive parameters of female lesser prairie-chickens (black =yearling; white
= adult) from 1998 to 2002. Note the consistency among clutch size, incubation date, and probability of (re)nesting (A, B, and D) with only slight separation in nesting success (C) between the age-classes...……………………………………103
Chapter 4 Fig. 1. Life-cycle diagram and matrix for a 2 age-class pre-breeding model of female
lesser prairie-chickens. Notations for vital rates are defined in the text.…………142 Fig. 2. Results of the life-table response experiment (LTRE) show the contribution of
each vital rate to the difference in λ between Area I (lower half of panel) and Area II (upper half of panel). This retrospective analysis determined that advantages of NEST1a and CHICK on Area II and I, respectively, had the largest contribution to population growth. Notations for vital rates are defined in the text.……………..144
Chapter 5 Fig. 1. Mean canonical variates (CAN-1 and CAN-2), SEs (in both x,y) and their
respective 95 % confidence circles for lesser prairie-chicken use, non-use and absent sites……………..………………………….….……………………………..…....166
Fig. 2. Odds ratios (95% confidence limits) for roads occurring in lesser prairie-chicken
monthly-ranges in 2000 (A), 2001 (B), and 2002 (C). The dashed line indicates odds of 1 and confidence limits intersecting this line indicate odds not different than expected by chance………………………….….……………………………..…..168
Fig. 3. Odds ratios (95% confidence limits) for powerlines occurring in lesser prairie-
chicken monthly-ranges in 2000 (A), 2001 (B), and 2002 (C). The dashed line indicates odds of 1 and confidence limits intersecting this line indicate odds not different than expected by chance………………………….….…………………..170
Fig. 4. Differences in the mean number (95% confidence limits) of wells ha-1,000 between
monthly- and non-ranges as determined from Poisson rate regression. The dashed line indicates a difference of 0 and confidence limits intersecting this line indicate a observed values did not differ from those expected by chance.…………………..172
v
Appendix I Fig. 1.Sampling locations (black dots) and counties (gray polygons) of lesser prairie-
chickens for genetic evaluations. Oklahoma and New Mexico sites are from Van den Bussche et al. (2003)..…………………………………………….………..193
Fig. 2. Frequency of 45 haplotypes (pie charts) by geographic region and sampling
sites..………………………………………………………………………..…..195 Fig. 3. Neighbor-joining tree based on Nei’s unbiased minimum distance. Note the
geographic structuring of populations reflected in Fig. 2..……………………..197 Appendix II No figures
vi
LIST OF TABLES Chapter 1 Table 1. Numbers of adult and yearling lesser prairie-chicken males captured for the first
time (F) or recaptured (R) from a previous year, and trapping efforts in Finney County, Kansas……………………………………………..……………………38
Table 2. Age-specific mark-recapture modeling for male lesser prairie-chickens in Finney
County, Kansas, 1998-2002..…………………………………………..…..……39 Table 3. Local survival rates and recapture probabilities for male lesser prairie-chickens
separating survival immediately after first capture from later transitions, Finney County, Kansas, 1998-2002.………………………………………………..……40
Table 4. The numbers of within and between year interlek movements of banded male
lesser prairie-chickens, Finney County, KS, 1998-2002……………….…...…..41
Table 5. Estimates of survival for North American prairie grouse (Tympanuchus spp.) obtained from live encounters (B), dead recovery (R), known-fate telemetry (T), and age-ratios from wing (W) analyses…………………………………….……42
Table 6. Estimates for age-specific survival and breeding for male grouse of the world
obtained from live encounters (B), dead recovery (R), known-fate telemetry (T), and age-ratios from wing (W) analyses.…………………………………..…….43
Chapter 2 Table 1. Candidate live-dead recovery models for estimating survival of radiomarked
and banded lesser prairie-chickens in Finney County, Kansas, 1998-2000…...61 Table 2. Parameter estimates (± SE) for live-dead recovery model (S group+ handle,
pgroup+t, rgroup, F1) of radiomarked and banded male lesser prairie-chickens in Finney County, Kansas, 1998-2000..……….....……………………….…….…..62
Chapter 3 Table 1. Numbers of yearling (Y) and adult (A) lesser prairie-chickens radiomarked in
Finney County, Kansas 1997-2002….....………………………………………..88
Table 2. Candidate models and model statistics for summer survival (Apr-Nov) sex and age-specific lesser prairie-chickens in Finney County, Kansas, 1997-2002…….89 Table 3. Summer survival estimates for radiomarked lesser prairie-chickens after marking
(Apr – Nov) 1997-2002.….....…………………………………………………..90
vii
Chapter 3 (continued) Table 4. Candidate models and model statistics for seasonal (Apr-Mar) sex and age- specific lesser prairie-chickens in Finney County, Kansas, 2000-2003.………...91 Table 5. Seasonal survival estimates for radiomarked lesser prairie-chickens 12-months
after marking (Apr – Mar) 2000-2003..….....…………………...……..………...92
Table 6. Numbers and percentages of potential mortality causes of lesser prairie-chickens in Finney County, Kansas, 1997 – 2003..….....…………………...………..…...93
Chapter 4 Table 1. Habitat and landscape features for study areas I and II in southwestern Kansas,
1998-2003. Parameter and standard errors (in parentheses) presented when a variable was not directly measured...….....…………………...………..…….....133
Table 2. Parameters for demographic model of female lesser prairie-chickens. Parameters describe events occurring within each time step (1 Apr in year t to 1 Apr in year t + 1)....…...………………….…………………...………..…….....134
Table 3. Parameter estimates (θ) of vital rates of yearling (Y) and adult (A) lesser prairie-
chickens radio-marked in Finney County, Kansas 1998-2002..………..……....135
Table 4. Analytic elasticities and variance scaled (arc-sin) sensitivities (VSS) for lower- level vital rates of matrix.………………….………………...………..…….....136
Table 5. Results of a life stage simulation anlysis (LSA) in which each vital rate was
increased by 10, 20, and 30 % (of its current estimate) and its variability decreased by 10%. The proportion of simulated matrices (n = 1,000) for each vital rate and percent increase that resulted in λ ≥ 1, is a measure of management effectiveness……………..………………….………………...………..…….....137
Table 6. Results of a modified LSA in which each vital rate was targeted for a set rate of
30 and 40% for nesting and chick survival, and 50 and 60% for post-brood survival and hen survival. Variability was simultaneously decreased by ~20% for each rate. The proportion of simulated matrices (n = 1,000) for each vital rate and it management goal that resulted in λ ≥ 1, is a measure of management effectiveness……………..………………….………………...………..…….....138
Table 7. Ranks of upper-level elasticities from matrix elements for 4 grouse species. A
rank of 1 indicates the most important matrix element………...………..…...…139
Table 8. Ranks of lower-level elasticities from matrix elements for 4 grouse species. A rank of 1 indicates the most important vital rate.………...…………..…..…......140
viii
Chapter 5 Table 1. Parameter estimates of habitat and landscape features for lesser prairie-chicken
use, non-use and absent sites. Means and standard errors are presented for measured variables along with the standardized canonical scores (CAN-1 and CAN-2) for each variable. Columns with different letters indicate a significant difference (P < 0.05).……………..…………………….……………………...162
Table 2. Canonical scores and sagebrush density plants ha-1 ( ± SEs) of lesser prairie- chicken use area type. Differences in usage area type are defined in he text..…163
Table 3. Count data of landscape features for contingency table and Poisson rate regression modeling of lesser prairie-chicken monthly-ranges…………….…..164
Appendix I Table 1. Descriptive statistics based on DNA sequence data of a portion of the mtDNA
D-loop for 8 lesser prairie-chicken populations.………………………………..189 Table 2. Significance (P < 0.0017) of pairwise FST tests for mtDNA sequencing data from
8 populations of lesser prairie-chickens in 4 states. Pairs of populations significantly different are shown by + and those not significantly different are shown by –.…..…………..…………………….………………..……………...190
Table 3. Number of alleles and average expected and observed heterozygosity at 5
microsatellite loci for 3 populations of lesser prairie-chickens in southwestern Kansas…..…………..…………………….…………………….……………...191
Appendix II Appendix 2A. Morphometrics (mean ± SD) of yearling and adult male and female lesser
prairie-chickens Finney County, Kansas 2002………………………….….…..198 Appendix 2B. Morphometrics (mean ± SD) of male lesser prairie-chickens captured in
Comanche County, Kansas and Prowers County, Colorado 2002.………..……199
ix
ACKNOWLEDGEMENTS
Funding for this research was supported by Kansas Federal Aid in Wildlife
Restoration Projects W-47-R and W-53-R through Kansas Department of Wildlife and
Parks. The Division of Biology at Kansas State University provided logistical and
additional financial support through a teaching assistant stipend. Westar Energy Inc.,
provided funding for radiotelemetry equipment for the marking and tracking of lesser
prairie-chicken chicks. This project would not have occurred without the support of
private landowners throughout southwestern Kansas. Ed Neidert of Ulysses, Kansas is
thanked for safe and entertaining flights while conducting aerial telemetry.
This project was a collaborative effort every inch of the way, and has included a
variety of individuals from across the prairies. I thank James C. Pitman for his
determination and dedication to this project, and teaching me a thing or two about
hunting Meleagris gallopavo. G. Curran Salter was invaluable in the field (and in the
kitchen) since 1997, without his efforts the second portion of this work would not have
been nearly as successful. I also thank him for an introduction to the finer points about
shotguns. Brent E. Jamison provided solid insight in the initial stages of my research
program as to the important questions that needed asking, and later help discover “deep-
fried” North Dakota sharp-tails. The aforementioned “chicken-boys” were instrumental
in the success of this project, we shared a lot over the last 4 years: memories that will not
soon be forgotten.
I thank T. J. Whyte, and C. C. Griffin for assistance with trapping and tracking.
T. Fields (CSU), M. Bain (FHSU), C. Swank (KDWP), R. Rogers (KDWP), J. Yost
(CDOW), W. Bryant (CDOW), and K. Giesen (CDOW) have provided assistance in the
collection of blood samples from across the range for genetic analyses that stemmed from
x
xi
my work. I thank Dr. Sara J. Oyler-McCance (USGS) for selflessly tackling the genetic
analyses, her enthusiasm for this work (and her patience with a “non-geneticist”) has
been greatly appreciated. I would like to thank Randy D. Rogers (KDWP) for
stimulating discussions on prairie chickens, and willingness to share his knowledge of the
ever-changing prairie.
The members of my research committee are acknowledged for their advice and
support during this endeavor. I thank my major advisor, Dr. Robert J. Robel, for the
opportunity to take part in this project and allowing me to ask my own questions. I thank
Dr. Brett K. Sandercock for blindly agreeing to serve on my committee before he left
Canada and arriving at KSU. His knowledge and skills in demographic analyses have
been invaluable to this project. Dr. Tom M. Loughin has provided substantial statistical
support over the course of this research, his sense of humor and willingness to share his
knowledge are appreciated. Dr. David A. Rintoul provided sound advice on my program
of study and useful critiques on my written work. Roger D. Applegate was the liaison
with KDWP his ability to provide vehicle support and additional funding was much
appreciated.
Above all, I thank my wife, Lisa. I Alcazar-Hagen, for her unwavering support of
my dedication to game bird research. Her ability to raise our sons, Dimitri and Sawyer,
in a positive and healthy way in my long absences is mystifying. I thank “our boys” for
tolerating my coming and going (mostly going) and reminding me about the important
things in life. I thank my parents William A. and Judith E. Hagen for their continuous
support and encouragement over the years. My extended Colorado family has been
supportive in many, many ways. Dr. David M. Armstrong also provided an editorial
review of this document, Thanks!
INTRODUCTION
The lesser prairie-chicken (Tympanuchus pallidicinctus) is one of 3 prairie grouse
in the genus Tympanuchus. Similar to its congeners, the greater prairie-chicken (T.
cupido) and sharp-tailed grouse (T. phasianellus), the lesser prairie-chicken requires
native grasslands or shrublands for breeding, nesting, brood rearing and survival, but has
one of the most restricted ranges of North American grouse (Giesen 1998). The lesser
prairie-chicken occupies mixed-grass and shrub prairies in Colorado, Kansas, New
Mexico, Oklahoma, and Texas. The sand sagebrush (Artemisia filifolia) prairies
consisting of bluestem (Agropogon spp.) and dropseed grasses (Sporobolus spp.) occur in
the sand hills of Colorado, Kansas (south of the Arkansas River), the northeastern
panhandle of Texas, and Oklahoma. The sand shinnery oak (Quercus havardii) savannas
occur in southeastern New Mexico, and the southwestern panhandle of Texas. The
mixed-grass prairie with little or no shrub component occurs north of the Arkansas River
in Kansas.
Although lesser prairie-chicken range was limited historically, human
disturbances have further reduced its distribution and abundance (Giesen 1998,
Woodward et al. 2001, Fuhlendorf et al. 2002). The continued destruction and
degradation of the mixed-grass prairies has been the primary cause of the population
declines since the turn of the 20th century (Taylor and Guthery 1980). It is hypothesized
that lesser prairie-chicken habitat has sustained a loss of >97% and population abundance
and distribution paralleled those losses (Taylor and Guthery 1980). Much of the early
habitat loss was attributed to extensive dry-land agriculture throughout the mixed-grass
prairies, followed by the Dust Bowl of the 1930s. During the 1960s the development of
1
center-pivot irrigation systems revolutionized agriculture in the semi-arid regions of the
southern Great Plains. This initiated a second wave of habitat loss and fragmentation that
lasted until the 1980s (Waddell and Hanzlick 1978). Although such large-scale
conversions of lesser prairie-chicken habitat have slowed, insidious changes to the
landscape have taken their toll (Woodward et al. 2001). Urbanization and suburban
housing developments, energy development, and invasion of woody species such as
eastern red cedar (Juniperus virginiana), osage orange (Maclura pomifera), and mesquite
(Prosopis spp.), are all sources of habitat loss or degradation.
Despite the general slowing trend in habitat loss, lesser prairie-chicken
populations have continued to decline range-wide (Braun 2000). Given its restricted
range, and isolated and unstable populations, the lesser prairie-chicken was petitioned for
listing as threatened under the Endangered Species Act in 1995 (U.S. Fish Wildlife
Service 2002). The U.S. Fish and Wildlife Service (USFWS) determined that such a
classification was “warranted but precluded” due to other species priorities. Currently
the priority for listing the lesser prairie-chicken is reviewed annually by the USFWS.
The conservation concern for the lesser prairie-chicken has warranted a
substantial body of both applied and basic research. In this introduction, I will outline
historical distribution and abundance of lesser prairie-chickens in Kansas, provide insight
to some of the knowledge gaps for the species, and indicate how I addressed those in the
following chapters of my dissertation.
2
LESSERS PRAIRIE CHICKENS IN KANSAS
Past distribution
The first studies of lesser prairie-chickens in Kansas (Baker 1953, Schwilling
1955) documented the species in 14 counties throughout sand sagebrush prairies,
generally south of the Arkansas River (Fig. 1). Baker (1953) confirmed a record of a
lesser prairie-chicken taken in January 1921 from Logan County in the northwestern
portion of the range (Fig. 1). Baker (1953) also noted a correspondence with a state game
protector (SGP) of Kansas Department of Wildlife Parks (KDWP) indicating that lesser
prairie-chickens once occupied counties as far north as Ellis, Ness, and Graham.
Schwilling (1955) documented 2 relatively small populations of birds at the periphery of
the known range in 1955: 1) on the border of Scott, Lane and Finney counties, and 2)
along the border of Finney and Hodgeman counties (Fig. 1). White (1963) reported
prairie chickens of unknown species in Ness and Hodgeman counties (Fig. 1). Waddell
(1977) summarized rural mail carrier survey observations of prairie chickens from 1962
to 1976, and found prairie chickens (species not identified) as far north as Ellis and
Russell counties. Observations in eastern counties included: Ellsworth, Rice, Reno, and
Harper (Fig. 1). Horak (1985) surveyed district biologists and SGPs to identify the
distribution of both prairie chicken species in 1980. Notably biologists and SGPs from
Region 6 (i.e., mixed prairies north of the Arkansas River) indicated the presence of 10%
lesser prairie-chickens, 10% unknown species, and 80% greater prairie-chickens (Horak
1985). Region 1 (the core of the lesser prairie-chicken range) was reported as having 5%
unknown species apparently in Hodgeman County. While inconclusive, these reports
combined suggest that the lesser prairie-chicken has occupied various habitats throughout
3
the western portion of the state over the last 100 years. If this distribution has fluctuated
naturally through time, it is likely dependent upon changes in the moisture gradient and
changes in land use (Baker 1953).
Present distribution
Currently lesser prairie-chickens occupy much of the range described above
although the abundances have declined (Fig. 1). In 1997, it was reported that substantial
numbers of lesser prairie-chickens were documented north of the Arkansas River in
northeastern Finney and western Hodgeman counties (KDWP unpublished data). As the
effort increased to document lesser prairie-chickens north of the Arkansas River more
populations were identified (KDWP unpublished data). Notably 4 counties have first
reported records of lesser prairie-chickens, Gove, Greeley, Wallace, and Wichita (Fig. 1).
Currently lesser prairie-chickens have been documented from listening surveys in 16
counties north of the Arkansas River. Although 10 established KDWP survey routes
south of the Arkansas River have indicated long-term declines, it is less clear as to the
long-term trends in the northern portion of the state. Lesser prairie-chickens currently
occupy 31 of 39 counties estimated to have been occupied historically (Jensen et al.
2000).
Abundance
Lek survey data (1964 -2002) from 10 survey routes indicated that over the long-
term, population indices have declined (Fig. 2). Jensen et al. (2000) quantified this
relationship through 1999 and indicated that year explained little of the variation in the
data, but the declining slopes were significantly different from zero. I examined the
relationship between lek survey data (number of birds/mi2) qualitatively and the amount
4
of pastureland available in each county where survey routes occurred. As an index to
available rangeland for prairie chickens I used U.S. Census of Agriculture data (1964-
1997) on acreage of pastureland. These data were collected at approximately 5-year
intervals; thus I averaged lek count data as bird/mi2 for the interval between censuses. I
used English units of measure to maintain consistency with KDWP databases. I assumed
that land use had not changed since 1997 and used the same pastureland values for 2002.
The resulting plot (Fig. 2) indicated a threshold effect where prairie chicken indices
increased as some portion of land was converted to agriculture but crashed in the early
1980s. Although the amount of pastureland in these survey routes has changed little in
the past 20 years, populations have continued to decline. Jamison (2000) found a
significant non-linear relationship for the Finney County survey data as it related to the
number of irrigation well permits issued in a given year. Thus, both of these independent
measures of land use suggest potential long-term negative impacts of large-scale
conversions of native rangeland to agricultural land.
KNOWLEDGE GAPS
The “warranted but precluded” status listing of the lesser prairie-chicken has
initiated a significant amount of research examining both intrinsic as well as extrinsic
factors that may be contributing to population trends. Surveys of disease (Hagen et al.
2001, Peterson et al. 2002, Wiedenfeld et al. 2002), helminthic parasites (Robel et al. in
press), and genetics (Van den Bussche et al. 2002), generally, have reported that these
intrinsic factors were not limiting the populations. Alternatively, Woodward et al. (2001)
identified landscape characteristics that were correlated with declining and stable
populations. Generally, populations occupying landscapes with slow rates of shrubland
5
habitat loss, were more stable than populations in landscapes with faster rates of habitat
loss. Most of this shrubland loss had been to either woody encroachment or suburban
development (Woodward et al. 2001). However, these changes in landscape were not
linked with demographic rates. Jamison (2000) suggested that depressed demographic
rates such as nest success and chick survival were the primary limiting factors in
southwestern Kansas. Jamison (2000) also examined survival rates and probable causes
of mortality of radiomarked males and females for 6-month periods (Apr-Sep). From
Jamison’s (2000) work it was suggested that adult survivorship during summer and early
fall was not a limiting factor. Winter survival was estimated for 1 year for males and
suggested that it was not less than that of summer.
Few data exist on the demography of the lesser prairie-chicken. Although short-
term studies (≤ 3 years) have quantified nest success (Riley et al. 1992, Giesen 1994) and
annual survival (Jamison 2000), only 1 study (Campbell 1972) estimated survival from
10 years of band recoveries. No studies have taken a comprehensive approach to
analyzing the demography and life-history strategy of the lesser prairie-chicken.
Evaluating age-specific variation in survival has been a central tenet of avian
population biology (Martin 1996). Such information is necessary for understanding
population dynamics and life-history strategy. Age-specific susceptibility to predation
and harvest has been of particular interest to researchers working on species of
conservation concern. The timing of mortality and annual survival are important
demographic rates in the study of life history, mating systems, and management actions
(Caizergues and Ellison 1997). Understanding the annual variation in the timing of
mortality and its severity are especially important for grouse species management
6
(Bergerud 1988), as females in promiscuous mating systems (unipaternal care) may have
an increased mortality risk during incubation and brood-rearing periods. Age-specific
survival has not been previously examined in the lesser prairie-chicken, and may yield
important insights to population dynamics.
Age-specific survival and fecundity are important vital rates as inputs to
population viability and sensitivity models. Previous demographic modeling of greater
and Attwater’s prairie-chicken (T. c. attwateri) indicated that nesting, chick survival, and
post-brood survival (i.e., survival from fledging to first breeding) had the highest
elasticities, and would be predicted to have the greatest effect on population growth
(Wisdom and Mills 1997, Peterson et al. 1998). This coincides with Bergerud’s (1988)
assessment of all grouse species that recruitment is the driving force behind population
cycling and persistence. Despite the tenuous status of the lesser prairie-chicken, no
comprehensive demographic analyses have been conducted, because annual and/or long-
term data sets are limited on this species.
Lesser prairie-chicken habitat use has been examined at the landscape (Jamison
2000, Woodward et al. 2001, Fuhlendorf et al. 2002), and microhabitat scale (i.e.,
vegetation characteristics at nests or use sites). While both types of studies have
demonstrated prairie chicken affinity for vegetation classes and structure, none have
examined habitat use as it pertains to human structures within a given landscape (i.e.,
meso-scale use). Understanding what features within a habitat patch make it less
suitable is critical to future assessment and conservation planning for lesser prairie-
chickens.
7
The primary objective of my dissertation was to synthesize and examine the
sensitivity of lesser prairie-chicken demography to changes in various vital rates. Vital
rates identified as having the largest impact on population growth rates should be
priorities for conservation efforts (Caswell 2001). I compared the demography of 2 lesser
prairie-chicken populations in southwestern Kansas.
This study is the culmination of 2 Federal Aid in Wildlife Restoration projects,
W-47-R and W-53-R. Field work for W-47-R began in spring 1997 and was completed
in November 1999, and W-53-R was initiated in December 1999 and concluded in March
2003. The objectives of this research were to 1) examine the age-specific survival rates
of male lesser prairie-chickens from mark-recapture data, 2) examine the potential biases
in survival estimates of radiomarked males, 3) examine known-fate survival rates for 9-
month period (1997-2002) and the timing of mortality for males and females, 4) examine
known-fate survival rates for a 12-month period (2000-2003) and timing of mortality for
males and females, 5) identify the probable cause of mortality in radiomarked birds, 6)
determine the stability of population growth using Leslie matrix models, 7) identify vital
rates to which population growth rate was most sensitive, and 8) model the probability of
within patch habitat use based on anthropogenic features on the landscape. This
dissertation is presented in 5 self-contained chapters written in the style of the Journal of
Wildlife Management.
LITERATURE CITED
Baker, M. F. 1953. Prairie chickens of Kansas. State Biological Survey, University of
Kansas. Lawrence, Kansas, USA.
Bergerud, A. T. 1988. Population ecology of North American grouse. Pages 578-685 in
A. T. Bergerud and M. W. Gratson, editors, Adaptive Strategies and Population
8
Ecology of Northern Grouse. Wildlife Management Institute, Washington D.C.,
USA.
Braun, C. E. 2000. Preface: special section of Prairie Naturalist on the lesser prairie-
chicken. Prairie Naturalist 32: 133-134.
Caizerigues, A., and L. N. Ellison. 1997. Survival of black grouse Tetrao tetrix in the
French Alps. Wildlife Biology 3: 177-186.
Campbell, H. 1972. A population study of lesser prairie-chickens in New Mexico. Journal
of Wildlife Management 36: 689-699.
Caswell, H. 2001. Matrix population models. Sinauer. Sunderland, Massachusetts, USA.
Fuhlendorf, S. D., A. J. Woodward, D. M. Leslie, and J. Shackford. 2002. Multi-scale
effects of habitat loss and fragmentation on lesser prairie-chicken populations in
the US Southern Great Plains. Landscape Ecology 17: 617-628.
Giesen, K. M. 1998. Lesser Prairie-Chicken (Tymapanuchus pallidinctus). F. Gill and A.
Poole, editors. The Birds of North America. No. 364. The Birds of North America
Inc., Philadelphia, Pennsylvania, USA.
Giesen, K. M. 1994. Movements and nesting habitat of lesser prairie-chicken hens in
Colorado. The Southwestern Naturalist 39: 96-98.
Hagen, C. A., S. S. Crupper, R. D. Applegate, and R. J. Robel. 2002. Prevalence of
Mycoplasma antibodies in lesser prairie-chicken sera. Avian Diseases 46: 708-
712.
Horak, G. J. 1985. Kansas prairie-chickens. Kansas Fish and Game Commission, Pratt,
Kansas, USA.
Jamison, B. E. 2000. Lesser prairie-chicken chick survival, adult survival, and habitat
selection and movement of males in fragmented landscapes of southwestern
Kansas. Thesis, Kansas State University, Manhattan, Kansas, USA.
9
Jensen, W. E., D. A. Robinson, and R. D. Applegate. 2000. Distribution and population
trends of lesser prairie-chicken in Kansas. Prairie Naturalist 32: 169-176.
Martin, K. 1995. Patterns and mechanisms for age-dependent reproduction and survival
in birds. American Zoologist 35: 340-348.
Peterson, M. J., W. T. Grant, and N. J. Silvy. 1998. Simulation of reproductive stages
limiting productivity of the endangered Attwater's prairie-chicken. Ecological
Modelling 111: 283-295.
Riley, T. Z., C. A. Davis, M. Oritz, and M. J. Wisdom. 1992. Vegetative characteristics
of successful and unsuccessful nests of lesser prairie-chicken. Journal of Wildlife
Management 56: 383-387.
Robel, R. J., T. L. Walker Jr., C. A. Hagen, R. K. Ridley, K. E. Kemp, and R. D.
Applegate. 2003. Helminth parasites of the lesser prairie-chicken in southwestern
Kansas: incidence, burdens, and effects. Wildlife Biology 8: in press.
Schwilling, M. D. 1955. A study of the lesser prairie-chicken in Kansas. Kansas Forestry,
Fish and Game Commission, Pratt, Kansas, USA.
Taylor, M. A., and F. S. Guthery. 1980. Status, ecology, and managment of the lesser
prairie-chicken. USDA Forest Service. General Technical Report RM-77. Fort
Collins, Colorado, USA.
US Department of the Interior Fish and Wildlife Service. 2002. Endangered and
threatened wildlife and plants; review of species that are candidates or proposed
for listing as endangered or threatened: the lesser prairie-chicken. Federal
Register 67: 40667.
Van den Bussche, R. A., S. R. Hoofer, D. A. Wiedenfeld, D. H. Wolfe, and S. K.
Sherrod. 2003. Genetic variation within and among fragmented populations of
lesser prairie-chickens (Tympanuchus pallidicinctus). Molecular Ecology 12: 675-
683.
10
Waddell, B., and B. Hanzlick. 1978. The vanishing sandsage prairie. Kansas Fish and
Game Magazine1-4.
Waddell, B. H. 1977. Lesser prairie chicken investigations - current status evaluation.
Kansas Forestry, Fish and Game Commission, Pratt, Kansas, USA.
White, C. 1963. Distribution of prairie chicken. Kansas Forestry Fish and Game
Commission. W-23-R-1 Job No. D-1, Pratt, Kansas, USA.
Wiedenfeld, D. A., D. H. Wolfe, J. E. Toepfer, L. M. Mechlin, R. D. Applegate, and S.
K. Sherrod. 2002. Survey for reticuloendotheliosis viruses in wild populations of
greater and lesser prairie-chickens. Wilson Bulletin 114: 142-144.
Wisdom, M. J., and L. S. Mills. 1998. Sensitivity analysis to guide population recovery:
prairie-chicken as an example. Journal of Wildlife Management 61: 302-312.
Woodward, A. J., S. D. Fuhlendorf, D. M. Leslie, and J. Shackford. 2001. Influence of
landscape composition and change on lesser prairie-chicken (Tympanuchus
pallidinctus) populations. American Midland Naturalist 145: 261-274.
11
Fig. 1. Historic and current distribution of lesser prairie-chickens in Kansas. Note extra-
limital records for several studies denoted by small symbols (see map legend for details).
12
Fig. 2. Population trends (birds/mi2) of lesser prairie-chickens from lek survey data
(1964-2002) from 4 to10 survey routes (A). The relationship between lek survey data
and available rangeland for counties where surveys were conducted (B). Five-year
intervals were used because U.S. Census of Agriculture quantified available pasture at
those approximate intervals. Lek survey data were averaged across each 5-year interval
to illustrate the potential long term effects of large-scale habitat loss.
14
1965 1968 1973 1975 1978 1983 1985 1988 1993 1995 1998 20031970 1980 1990 2000
Avg.
Num
ber o
f bird
s m
i2
0
2
4
6
8
10
12
14
16
18
1964 1969 1974 1978 1982 1987 1992 1997 2002
Bird
s/m
i2
0
2
4
6
8
10
12
14
16
18
20
Avai
labl
e ra
ngel
and
(10,
000
ac)
20
30
40
50
60
70BirdsRange
(A)
(B)
15
CHAPTER 1
AGE-SPECIFIC SURVIVAL IN MALE LESSER PRAIRIE-CHICKENS
Abstract: Lesser prairie-chicken (Tympanuchus pallidicinctus) habitat and populations
have been reduced by more than 90% since the turn of the 20th Century. To aid in the
conservation of this “warranted but precluded” threatened species, more information is
needed on its basic ecology. Few data exist on the demography of the lesser prairie-
chicken. Evaluating age-specific variation in survival has been a central tenet of avian
population biology. I examined the hypothesis of delayed reproductive effort and
increased survival of the yearlings in the lesser prairie-chicken. I used age-structured
mark-recapture models to estimate the local survival rates of banded yearling and adult
male lesser prairie-chickens from mark-recapture data, and used inter-lek movements as
an index of site fidelity in southwestern Kansas. I compared these survival rates to other
grouse species and mating systems of the holarctic region. I modeled age-structured local
survival (φ) and recapture (p) probabilities using extensions of the Cormack-Jolly-Seber
approach for open populations. Three hundred and seventy-six male prairie-chickens
(173 yearlings, 203 adults) were captured from 1998-2002. Local survival rates of male
lesser prairie-chickens were ranked: yearling (φ1 = 0.615, SE = 0.068) > adult (φ1 =
0.485, SE = 0.058) > older adults (φ2 = 0.347, SE = 0.047). Twenty percent of
recaptured yearlings (n = 60) switched leks in their second year, and their odds of
switching were 2.5 times as high as those adults (8%, n = 65). Lesser prairie-chickens in
southwestern Kansas fit models of bimaturism in survival and breeding effort.
16
INTRODUCTION
Lesser prairie-chicken (Tympanuchus pallidicinctus) habitat and populations have
been reduced by more than 90% since the turn of the 20th Century (Giesen 1998). To aid
in the conservation of this “warranted but precluded” threatened species (US Department
of Interior, Fish and Wildlife Service 2002), more information is needed on the basic
ecology of this prairie grouse. Few data exist on the demography of the lesser prairie-
chicken. Although, short-term studies (≤ 3 years) have quantified nest success (Riley et
al. 1992, Giesen 1994) and annual survival (Jamison 2000), only 1 previous study
(Campbell 1972) estimated survival from 10 years of band recoveries. No studies to date
have examined age-specific rates in survival.
Robust estimates of annual survival are useful for 2 reasons: understanding
management efforts and basic science. Survival is one of several demographic rates that
can affect fluctuations in population numbers of grouse (Bergerud 1988, Martin 1996).
Additionally, age-specific survival and reproductive rates may covary with the type of
mating system in grouse (Wiley 1974). Wiley (1974) hypothesized that the evolution of
promiscuous mating systems in grouse was a result of 4 concomitant factors: sexual
differences in the age at first breeding, sexual size dimorphism, a lack of a need for bi-
parental care for precocial young, and delayed age of first reproduction in males.
Foregoing reproduction in yearling males is a trade-off for increased survival in the short-
term. This suggests that there is a cost associated with reproductive effort in the
promiscuous mating systems of grouse. If true, the predicted pattern in survival for males
in promiscuous systems should exhibit a senescent pattern after the first year of
reproduction. This pattern may be particularly acute for lek mating species where male-
17
male competition is more intensive than other promiscuous systems (Bergerud 1988). (A
lek is defined as an arena used primarily for breeding, and other aspects of reproduction
and life requirements are independent of the lek). Alternatively, yearling males should
reproduce in monogamous systems, as there is a lower risk associated with reproductive
activity (Wiley 1974).
Few studies have directly examined age-specific variation in life history traits of
male grouse (Wittenberger 1978, Lewis and Zwickel 1982). Some studies have indirectly
examined these patterns by estimating annual survival of prairie grouse (Robel et al.
1972, Hamerstrom and Hamerstrom 1973, Burger et al. 1991, Schroeder 1997, Giesen
1999). Studies of other promiscuous grouse species have demonstrated age-specific
survival rates of males with yearling males surviving better than older birds (Braun 1979,
Lewis and Zwickel 1982, Angelstam 1984, Lewis and Jamieson 1987, Zablan et al.
2003). It has been suggested (Wiley 1974, Lewis and Zwickel 1982) that this aging
pattern of adult survival is a result of young males foregoing reproduction in the first year
and increasing their survival from a lack of reproductive effort (Emmons and Braun
1984).
Most studies on grouse survival have used return rates (i.e., the proportion of
marked individuals that are recaptured or resighted in a subsequent year) as an estimate
of survival. Interpretation of these estimates can be difficult because return rates are the
product of 4 probabilities: true survival (S), site fidelity (F), site propensity (γ*), and
detection (p*). Alternatively mark-recapture methods of live encounter data can yield
improved survival estimates based on return rates by estimating a local survival rate (φ =
S × F) corrected for the probability of recapture (p = p*γ*). Thus mark-recapture
18
statistics alone cannot estimate true survival, and ancillary data on F are needed to
identify potential biases in φ.
I examined Wiley’s (1974) hypothesis of delayed reproductive effort and
increased survival of yearlings in the lesser prairie-chicken. I used age-structured mark-
recapture models to estimate local survival rates of banded yearling and adult male lesser
prairie-chickens from mark-recapture data, and used inter-lek movements as an index to
site fidelity in southwestern Kansas. I compared these survival rates to other grouse
species and mating systems of the holarctic region.
METHODS
Study species
The lesser prairie-chicken has one of the most restricted ranges of North
American grouse (Giesen 1998), occupying mixed-grass and shrub prairies in Colorado,
Kansas, New Mexico, Oklahoma, and Texas. Although lesser prairie-chicken range was
limited historically, human disturbances have further reduced its distribution and
abundance (Giesen 1998, Woodward et al. 2001, Fuhlendorf et al. 2002). The continued
destruction and degradation of the mixed-grass prairies has been the primary cause of the
population declines since the turn of the 1900’s (Crawford 1974). Conversion of sand
sagebrush (Artemisia filifolia) habitat to intensive center-pivot agriculture in Kansas
ceased in the 1980’s, but lesser prairie-chicken populations have continued to decline
(Jamison 2000, Jensen et al. 2000). Thus, an examination of the demography of this
species is needed to clarify what may be causing the populations to decline in
southwestern Kansas.
19
Study areas
The study region was comprised of 2 fragments (~5000 ha each) of native sand
sagebrush (hereafter sandsage) prairie near Garden City, Finney County, Kansas (37° 52′
N, 100° 59′ W). Work was initiated on 1 area in 1998, and trapping and monitoring
efforts were expanded to include the second area in 2000. Prior to 1970, these 2 areas
were part of 1 contiguous track of native sandsage prairie. The development of center
pivot irrigation left these areas as 2 fragments with about 15 km of agricultural fields
between them (Waddell and Hanzlick 1978).
Capture and handling
Lesser prairie-chickens were captured during spring at 20 lek sites using walk-in
funnel traps (Haukos et al. 1991, Schroeder and Braun 1991). Traps were placed on all
known leks (> 3 displaying males) found in native prairie on the study sites. No attempt
was made to trap birds displaying on agricultural fields because these leks sites were
unstable. Traps were rotated among groups of 2–3 leks every 7–11 days (trap period x =
7.9, SD = 1.7), and each rotation was defined as a trap period. From 1998 to 1999, 2–3
leks were trapped simultaneously on each study area per trap period (3 or 4 periods per
spring), and in the 2000 to 2002 portion of the study, 4–6 leks were trapped per trap
period (Table 1). At first capture, birds were aged as yearling (~10 months of age) or
adult (≥ 22 months) based on shape, wear, and coloration of the ninth and tenth primaries
(Amman 1944, Copelin 1963). Yearlings were identified as birds with pointed and
frayed tips of the ninth and tenth primaries, and white spotting within 2.5 cm of the tip of
the tenth primary. Body mass was measured (± 2.5 grams) using a Pesola spring scale.
Birds were marked with serially numbered aluminum leg bands. Recoveries of banded
20
birds were few (n = 21) and precluded the use of joint live-encounter dead-recovery
models for estimating age-specific survival rates. Ancillary movement data were used
from radiomarked birds on the study areas (Jamison 2000, Hagen unpublished data) to
examine the potential effects of permanent emigration.
Survival analyses
I modeled age-structured local survival (φ) and recapture (p) probabilities using
extensions of the Cormack-Jolly-Seber approach for open populations, and following the
general procedures of Burnham and Anderson (1998). Mark-recapture analyses were
conducted in program MARK 3.0 (White and Burnham 1999) following general steps: 1)
selection of the global model, 2) goodness-of-fit tests (GOF), and 3) fitting and selection
of reduced models with fewer parameters. Subscripts following parameters indicate
explanatory variables (e.g., φ age*t describes a model that includes both age- [yearling or
adult as a group effect] and time-dependence in survival). Parameter superscripts “1” and
“2” denote the interval after initial release or all subsequent intervals, respectively (e.g.,
φ1age*t, φ2
t describes a model that features both age- and time-dependence in survival after
the initial release, and time-dependence in survival at all subsequent encounters). Lastly,
I use a subscript (c) to describe models with parameters held constant (e.g., φ c, describes
a model where local survival is held constant).
Global models included age and annual conditions because these factors affect
local survival in many other birds (Saether 1990, Martin 1996). Age at first capture was
treated as a group effect (e.g., group 1 = yearling, group 2 = adult). I modeled local
survival with a modified 2-age class model that separated the interval after first capture
(φ1) from all subsequent transitions (φ2). Thus the structure of my global model φ1age*t,
21
φ2t, pt allowed me to evaluate the effects of age on survival after the first capture
(Sandercock and Jamarillo 2002). All models were constructed the using design matrices
feature and the logit link function in program MARK 3.0.
Bootstrap GOF testing was used to test the 2 assumptions of equal capture and
survival probabilities. Model fit was calculated as the rank of the observed deviance
from the global model (φ1age*t, φ2
t, pt) relative to the bootstrap distribution (n = 1,000) of
model deviance (Cooch and White 2001). If the observed deviance value was > 95% of
the bootstrap data then the model fit was suspect, variances were adjusted for
overdispersion, and model selection was based on quasi-likelihood AICc
(QAICc)(Andersen et al. 1994). I used bootstrapped estimates of model deviance to
estimate the quasi-likelihood parameter or variance inflation factor (ĉ) and adjusted the
standard errors of parameter estimates (Cooch and White 2001). I calculated ĉ as the
observed deviance of the global model divided by the mean expected deviance from a
parametric bootstrapping (Cooch and White 2001). If ĉ was < 1, then ĉ was set to 1.
After examining the fit of the most general models and adjustments for
overdispersion were made, models with fewer parameters were fit to the data. I selected
the best-fit model based on the minimization of Akaike’s Information Criterion adjusted
for small sample sizes (AICc) and/or ĉ (QAICc) (Burnham and Anderson 1998). Models
with differences of AICc (∆AICc) values ≤ 2 from the best fit model were considered
equally parsimonious. The ratio of Akaike weights (wi / wj ) between 2 models was used
to quantify the relative degree that a pair of models was supported by the data (Burnham
and Anderson 1998).
22
The number of potential models was large, therefore I used a hierarchical
procedure to guide model fitting (Lebreton et al. 1992). Because my primary interest was
local survival (φ), recapture rate (p) was modeled first as a nuisance parameter. Local
survival was then modeled in 2 steps. First, the probabilities of local survival at later
transitions (φ2) were modeled followed by survival of the first interval (φ1) for the
modified-age model (Sandercock and Jamarillo 2002). Probabilities p and φ2 were
modeled either with time-dependence (t) or as a constant (c). Age and the age*time
interaction from these probabilities were excluded because I assumed that birds first
marked as yearlings would have similar survival and recapture rates as adults if they
returned after the first interval.
Parameter estimates were calculated from the best-fit model; where there was
more than 1 parsimonious model (∆AICc or ∆QAICc < 2), I used the model averaging
procedure in MARK. The model averaging procedure calculates a weighted average
parameter estimates for each transition from all models in the candidate set, weighting
estimates by using the Akaike weights (wi) specific to each model. I did not use the
variance components procedure due to the relatively short period of the study, thus
overall estimates of standard errors include both process and sampling variance.
Estimation of interlek movements
Local survival is a product of true survival and site fidelity. I aided my
interpretation of estimates using knowledge of site fidelity. Natal dispersal of yearling
male prairie chickens likely occurs in autumn (prior to my capture intervals), and
dispersal movement distances of males are usually localized (Bowman and Robel 1977,
Jamison 2000, Pitman 2003). Thus, the frequency of movements of banded yearling and
23
adult males were examined both within and between seasons to evaluate an overall
relationship between survivorship and fidelity. Previous work on other lek-mating grouse
has shown a greater frequency of movement between leks indicated a non-territorial
male, that was sampling leks for opportunities to reproduce in the future (Wiley 1974,
Emmons and Braun 1984). If on the first encounter in a successive year a bird was
recaptured on a lek other than the lek of first capture it was defined as a between-year
movement. If within a trapping season, a bird was recaptured on a lek other than first
capture only during the second encounter it was considered a within-year movement. I
use this conservative definition, because it is possible with a greater number of handlings
birds may have moved to avoid being handled in the future. A likelihood ratio test was
used to examine the propensity of yearlings to move between leks relative to that of
adults.
RESULTS
Three hundred and seventy-six male prairie-chickens (173 yearlings, 203 adults)
were captured from 1998–2002, and 150 males (78 yearlings, 72 adults) were recaptured
at least once (Table 1). Forty-six birds were treated as not released at last capture due to
known mortalities of radiomarked birds, removals for a parasite study, permanent
emigration, and trap mortalities.
Survival analyses
The parametric bootstrap GOF test indicated that the global model (φ1age*t, φ2
t, pt)
(P = 0.724) met the assumptions of mark-recapture analysis, and I did not need to adjust
for overdispersion because ĉ < 1. Thus, I used AICc for model selection. Modeling of
the recapture probabilities indicated that models with time-dependent and constant
24
recapture rates were equally parsimonious. Given that I had only 5 capture occasions,
capture effort per lek was similar in all years (Table 1), and model selection was identical
with pt but with one less estimable survival rate, all subsequent models were fit with pc.
Model selection based on AICc indicated that the most parsimonious model (φ 1age+t, φ 2c,
pc) with respect to survival was that which recognized differential survival in the age-
classes with an additive time effect (Table 2). Local survival rates of male lesser prairie-
chickens were ranked: yearling (φ1 = 0.615, SE = 0.068) > adult (φ1 = 0.485, SE = 0.058)
> older adults (φ2 = 0.347, SE = 0.047). Models with annual variation in survival of
older birds were not well supported by the data (∆AICc > 7), suggesting that survival at
later intervals is best understood as a constant rate.
Interlek movements
Twenty percent of recaptured yearlings (n = 60) switched leks in their second
year. Their odds of switching were 2.5 times as high (G = 4.735, df = 1, P = 0.030;
Table 4) as adults’ odds (8%, n = 65). Each age-class was approximately equally likely
(~17.5 %) to move between leks within a breeding season (G = 0.035, df = 1, P = 0.851;
Table 4). However, 4 and 15 % of yearlings and adults were recaptured > 3 times,
respectively (Table 4). This suggests that adults had a greater propensity to attend leks
(and presumably try to obtain copulations) than yearlings.
DISCUSSION
The major findings of this study were: 1) lesser prairie-chicken males had greater
survival in the first transition (yearlings > adults > older adults) than in later years, 2)
fidelity to lek sites was higher for adults than for yearlings, and 3) these patterns were
unlike most all other birds.
25
Survival models with additive annual variation in the yearling and adult age
classes were well supported by the data. This indicated that yearlings and younger adult
males (φ1) were susceptible to similar environmental factors, but older males (φ2) had a
consistently lower survival rate. However, this may have been tempered by the relatively
short period of my study, and a longer term may have yielded better resolution at these
intervals. Although Brown (1978) did not measure survival directly, he indicated that
harvest levels and yearling adult ratios of lesser prairie-chickens were positively
correlated with the previous year’s precipitation. Given the relatively short period of my
study, it is difficult to quantify what factors may have contributed to this annual variation
in survival.
Local survival (φ) has 2 components, true survival (S) and site fidelity (F). If age-
specific variation in φ was due to F, then permanent emigration should be greater in
adults. In fact, the interlek movement data demonstrate the opposite pattern, greater
movements among yearlings than adults. Therefore variation in φ is likely due to S, and
the observed patterns could be even more pronounced, because φ is more likely to be an
underestimate of S for yearlings.
Radiotelemetry data also indicated low rates of emigration, as only 1 yearling and
1 adult of 119 (1.7%) radiomarked males in this study were known to permanently
emigrate during the breeding season. Although, yearlings were more likely to switch leks
locally than adults were. The relatively high recapture rates (p = 0.73) in this study
appear not to have been biased by temporary emigration, as only 1 radiomarked adult
male was documented moving between my 2 study sites during the breeding season.
Two other males (1 yearling and 1 adult) moved between study sites after the breeding
26
season. The adult was recovered dead on the area to which he emigrated. The yearling’s
transmitter fell off several months after he emigrated but he was recaptured at his original
lek of capture the following spring.
Survival in grouse
My overall estimates of lesser prairie-chicken survival were slightly elevated
when compared to that of other banding studies of prairie grouse (Table 5). Given the
various methods used to estimate survival it is difficult to compare these rates directly. I
assumed that survival estimates based on return rates were biased low because recapture
probabilities were not estimated (Campbell 1972). The range of my estimates (0.234 –
0.792) encompassed most of those reported elsewhere (Table 5), and I conclude that
annual local survival of male lesser prairie-chickens is comparable to its congeners.
Evidence for differential survival of yearlings and adults is anecdotal in previous
work on prairie grouse (Campbell 1972, Hamerstrom and Hamerstrom 1973) (Table 6).
Life-table analyses of greater prairie-chickens (Tympanuchus cupido) in Wisconsin were
suggestive of age-specific survival (yearling = 0.50, adult = 0.48) but differences in
survival rates were < 5% (Hamerstrom and Hamerstrom 1973). Campbell’s (1972) data
from New Mexico indicated differential survival in male lesser prairie-chickens (yearling
= 0.35, adult = 0.30). Thus, age-specific patterns in survival exist but they have not been
examined with rigorous quantitative techniques. Wiley (1974) hypothesized that
promiscuity in grouse resulted from such age-specific patterns, because yearling male
grouse would forego reproduction in their first year, thereby enhancing their probability
of survival in the short-term. This prediction assumes a cost of reproduction. The
27
evidence for such costs would be supported by senescent age-patterns in males, and
should occur in other promiscuous grouse species.
These age-specific patterns in survival comes from the most sexually dimorphic
grouse species, the blue grouse (Dendragapus obscurus), sage-grouse (Centrocercus
spp.), capercaillie (Tetrao urogallus) and black grouse (Tetrao tetrix). Lewis and
Zwickel (1982) found that yearling blue grouse had increased survival to the next year
relative to older birds. Greater sage-grouse (C. urophasianus) males exhibit similar
survival patterns (Braun 1979, Zablan et al. 2003). Emmons and Braun (1984) reported
that all radiomarked yearling (n = 17) greater sage-grouse switched leks on average 2.8
times during the breeding season, presumably seeking open territories. This suggests that
yearlings spend most of the first spring sampling (Emmons and Braun 1984) and have
significantly greater survival than older birds (Braun 1979, Zablan et al. 2003). Data on
annual survival rates of capercaillie in these age-classes are sparse (Moss 1987). Moss
(1987) suggested that mortality of older birds was slightly (~10%) elevated as compared
to yearlings but was not statistically different. Angelstam (1984) reported that
radiomarked male black grouse yearlings all survived the breeding and summer season,
but older birds suffered higher mortality rates. How, then, does local survival relate to
reproductive effort and mating systems?
Little information is available on the breeding behavior of the lesser prairie-
chicken (Haukos and Smith 1999). However, from studies of its congeners it is well
established that the majority of copulations are performed by a few adult males (≥ 2 yrs
of age), and a smaller portion of copulations are secured by yearlings (Hamerstrom and
Hamerstrom 1973, Robel 1970, Gratson et al. 1991). Generally, prairie grouse males that
28
hold peripheral territories acquire fewer copulations than central males (Robel 1970,
Gratson et al. 1991). Robel et al. (1970) found that interlek movements of non-territory
holding adult greater prairie-chickens occurred at low frequencies, but were common
among yearlings. Sharp-tailed grouse (Tympanuchus phasianellus) males that held
peripheral territories (mostly yearlings) had slightly higher return rates (39 %) than those
of centrally located males (33 %) (Gratson et al. 1991). Gratson et al. (1991) also noted
the presence of a highly mobile subpopulation of ‘non-lekking’ males that was mostly
yearlings. From Robel et al. (1970) and Gratson et al. (1991) it can be suggested that the
frequency of movement between leks may serve as an index to breeding activity.
Although, the within season interlek movements of adults and yearlings in my study were
equivalent (~17 %), the annual switching of leks by yearlings (20 %) suggests that more
sampling was occurring than was detected, and it is possible that yearlings increase their
survival rate by reducing the time defending or acquiring territories. Thus, lesser prairie-
chicken males appear to fit Wiley’s (1974) hypothesis of delayed breeding and increased
survival.
Age-specific patterns in reproduction are most pronounced in the more sexually
dimorphic grouse species, blue grouse, sage-grouse, and capercaillie. Lewis and Zwickel
(1982) found that yearling blue grouse forego reproduction in the first year. Greater
sage-grouse exhibit similar survival (Braun 1979, Zablan et al. 2003) and reproductive
patterns (Wiley 1974, Hartzler and Jenni 1988). The delayed breeding tactics of yearling
capercaillie and sage-grouse are well known (Storch 2001), and males do not achieve
complete adult breeding plumage until ≥ 2 breeding seasons. This strategy presumably
provides for higher reproductive success as an older bird.
29
Wiley’s (1974) predicted patterns of survival and breeding effort in grouse were
supported by the relatively few studies that documented (or could be gleaned from
reported returns) age-specific survival and breeding effort (Table 6). Risk modeling of
the lek breeding great snipe (Gallinago media) indicated that individuals with low
reproductive probabilities spent more time hiding than individuals with higher
reproductive probabilities (Kalas et al. 1995). This translated into higher survival rates
for individuals not participating as frequently in breeding activities. Alternatively, long-
lived species such as long-tailed manakins (Chiroxiphia linearis) exhibit the opposite
pattern where the most successful breeders (birds > 5 yrs old) with high site fidelity had
the highest survival rates (McDonald 1993). However, manakins are much longer lived
resulting in a different life history strategy. The observed pattern of survival and fidelity
in my study correspond to a life-history strategy of a relatively short-lived promiscuous
species. The monogamous grouse species (Lagopus spp. and Bonasa bonasia) tend to
have similar survival rates between yearlings and older males, and both breed (Martin et
al. 2000, Montadart and Leonard 2002). This does not suggest that all yearlings breed
(Hannon and Smith 1984), but when territory holders were removed, yearlings occupied
vacant territories (Martin and Hannon 1987).
In some studies, the benefit of maintaining a territory determines over-winter
survival of males (Jenkins et al. 1963, Pedersen 1984). Although, fall territory ownership
is only directly related to over-winter survival of red grouse males (Lagopus lagopus
scoticus), territory ownership in general conveys higher fitness of other ptarmigan
species. Thus, it would appear that the benefits of reproducing in the first year outweigh
the potential costs to future survival in monogamous species.
30
In conclusion, I observed age-dependent effects for lek fidelity and local survival.
It was surprising that yearling lesser prairie-chicken males survived at a higher rate than
older birds; this result was contrary to studies of most other birds. The observed pattern
in survival rates were not likely biased by emigration rates, because the interlek
movement data demonstrated greater movements among yearling than adults. Therefore
variation in local survival is likely due to changes in true survival and the observed
patterns could be even more pronounced, because local survival is more likely to be an
underestimate of true survival for yearlings. These age-specific patterns are perhaps
more pronounced in lek mating grouse than other mating systems, where male-male
competition is reduced. Lesser prairie-chickens in southwest Kansas fit models of
bimaturism in survival and breeding effort, but further work is needed to model survival
with age and direct measures of reproductive effort as covariates.
ACKNOWLEDGMENTS
I thank J. O. Cattle Co., Sunflower Electric Corp., Brookover Cattle Co., and P. E.
Beach for property access. C. G. Griffin, B. E. Jamison, J. C. Pitman, G. C. Salter, T. G.
Shane, T. L. Walker, Jr., and T. J. Whyte assisted with trapping and banding of prairie
chickens. B. K. Sandercock was instrumental in providing advice on mark-recapture
analyses. Financial and logistical support was provided by Kansas Department of
Wildlife and Parks (Federal Aid in Wildlife restoration projects, W-47-R and W-53-R),
Kansas Agricultural Experiment Station (Contribution No. 03-_______), Division of
Biology at Kansas State University, and Westar Energy Inc.
LITERATURE CITED
Amman, G. A. 1944. Determining the age of pinnated and sharp-tailed grouse. Journal of
31
Wildlife Management 8: 170-171.
Amman, G. A. 1957. The prairie grouse of Michigan. Department of Conservation Game
Division, Lansing, Michigan, USA.
Andreev, A. V., F. Hafner, S. Klaus, and H. Gossow. 2001. Displaying behavior and
mating system in the Siberian spruce grouse (Falcipennis falcipennis Hartlaub
1855). Journal für Ornithologie 142: 404-424.
Angelstam, P. 1984. Sexual and seasonal differences in mortality of black grouse Tetrao
tetrix in boreal Sweden. Ornis Scandinavica 15: 123-134.
Bergerud, A. T. 1970. Population dynamics fo the willow ptarmigan Lagopus lagopus
alleni L. in New Foundland 1955 to 1965. Oikos 21: 299-325.
Bergerud, A. T. 1988. Population ecology of North American grouse. Pages 578-685 in
A. T. Bergerud and M. W. Gratson, editors, Adaptive Strategies and Population
Ecology of Northern Grouse. Wildlife Management Institute, Washington D.C.,
USA.
Boisvert, J. H. 2002. Ecology of Columbian sharp-tailed grouse associated with
Conservation Reserve Program and reclaimed surface mine lands in northwestern
Colorado. Thesis, University of Idaho, Moscow, Idaho, USA.
Bowman, T. J., and R. J. Robel. 1977. Brood break-up dispersal, mobility and mortality
of juvenile prairie-chickens. Journal of Wildlife Management 41: 27-34.
Braun, C. E. 1979. Evaluation of the effects of changes in hunting regulations on sage
grouse populations. Colorado Division of Wildlife. Progress Report W-37-R-32.
Denver, Colorado, USA.
Braun, C. E. 1969. Population dynamics, habitat, and movements of white-tailed
ptarmigan in Colorado. Dissertation, Colorado State University, Fort Collins,
Colorado, USA.
32
Brown, D. E. 1978. Grazing, grassland cover, and gamebirds. Transactions of the North
American Wildlife Conference 43: 477-485.
Burger, L. W., M. R. Ryan, D. P. Jones, and A. P. Wywialowski. 1991. Radio
transmitters bias estimation of movements and survival. Journal of Wildlife
Management 55: 693-697.
Burnham, K. P. and D. R. Anderson . 1998. Model selection and inference: a practical
information-theoretic approach. Springer. New York, New York, USA.
Campbell, H. 1972. A population study of lesser prairie-chickens in New Mexico. Journal
of Wildlife Management 36: 689-699.
Cooch, E., and G. C. White 2001. Using MARK-a gentle introduction.
http://www.cnr.colostate.edu/~gwhite/mark/mark.html (6 June 2002)
Copelin, F. F. 1963. The lesser prarie-chicken in Oklahoma. Oklahoma Wildlife
Conservation Department. Technical Bulletin No. 6. Oklahoma City, Oklahoma,
USA.
Emmons, S. R., and C. E. Braun. 1984. Lek attendance of male sage grouse. Journal of
Wildlife Management 48: 1023-1028.
Fuhlendorf, S. D., A. J. Woodward, D. M. Leslie, and J. Shackford. 2002. Multi-scale
effects of habitat loss and fragmentation on lesser prairie-chicken populations in
US Southern Great Plains. Landscape Ecology 17: 617-628.
Giesen, K. M. 1999. Columbian sharp-tailed grouse wing analysis: implications for
managment. Colorado Division of Wildlife. Special Report No. 74. Denver,
Colorado, USA.
Giesen, K. M. 1998. Lesser Prairie-Chicken (Tymapanuchus pallidinctus). F. Gill and A.
Poole, editors. No. 364. The Birds of North America Inc., Philadelphia,
Pennsylvania, USA.
33
Giesen, K. M. 1994. Movements and nesting habitat of lesser prairie-chicken hens in
Colorado. The Southwestern Naturalist 39: 96-98.
Gullion, G. W., and W. H. Marshall. 1968. Survival of ruffed grouse in a boreal forest.
The Living Bird 7: 117-167.
Hamerstrom, F. N., and F. Hamerstrom. 1973. The prairie chicken in Wisconsin.
Wisconsin Department of Natural Resources . Technical Bulletin No. 64.
Madison, Wisconsin, USA.
Hannon, S. J., and J. N. Smith. 1984. Factors influencing age-related reproductive
success in the willow ptarmigan. Auk 101: 848-854.
Haukos, D. A., L. M. Smith, and G. S. Broda. 1990. Spring trapping of lesser prairie-
chickens. Journal of Field Ornithology 61: 20-25.
Hulett, G. K., J. R. Tomelleri, and C. O. Hampton. 1988. Vegetation and flora of a
sandsage prairie site in Finney County, southwestern Kansas. Transactions of the
Kansas Academy of Science 91: 83-95.
Jamison, B. E. 2000. Lesser prairie-chicken chick survival, adult survival, and habitat
selection and movement of males in fragmented landscapes of southwestern
Kansas. Thesis, Kansas State University, Manhattan, Kansas, USA.
Jenkins, D., A. Watson, and G. R. Miller. 1963. Population studies on red grouse,
Lagopus lagopus scoticus (Lath.) in north-east Scotland. Journal of Animal
Ecology 32: 317-376.
Jensen, W. E., D. A. Robinson, and R. D. Applegate. 2000. Distribution and population
trends of lesser prairie-chicken in Kansas. Prairie Naturalist 32: 169-176.
Kalas, J. A., P. Fiske, and S. A. Saether. 1995. The effect of mating probability on risk
taking: an experimental study in the lekking great snipe. American Naturalist 146:
59-71.
34
Keppie, D. M. 1979. Dispersal, overwinter mortality, and recruitment of spruce grouse.
Journal of Wildlife Management 43: 717-727.
Kruijt, J. P., and J. A. Hogan. 1967. Social behavior on the lek in black grouse Lyrurus
tetrix tetrix (L.). Ardea 55: 203-240.
Lebreton, J.-D., K. P. Burnham, J. Clobert, and D. R. Anderson. 1992. Modeling survival
and testing biological hypotheses using marked animals: a unified approach with
case studies. Ecological Monographs 62: 67-118.
Lewis, R. A., and I. G. Jamieson. 1987. Delayed breeding in yearling blue grouse and
evaluation of two hypotheses. Condor 89: 182-185.
Lewis, R. A., and F. C. Zwickel. 1982. Survival and delayed breeding in male blue
grouse. Canadian Journal of Zoology 60: 1881-1884.
Martin, K. 1991. Experimental evaluation of age, body size, and experience in
determining territory ownership in willow ptarmigan. Canadian Journal of
Zoology 69: 1834-1841.
Martin, K. 1995. Patterns and mechanisms for age-dependent reproduction and survival
in birds. American Zoologist 35: 340-348.
Martin, K., and S. J. Hannon. 1987. Natal philopatry and recruitment of willow
ptarmigan in north central and northwestern Canada. Oecologia 71: 518-524.
Martin, K., P. B. Stacey, and C. E. Braun. 2000. Recruitment, dispersal, and
demographic rescue in spatially structured white-tailed ptarmigan populations.
Condor 102: 503-516.
McDonald, D. B. 1993. Demographic consequences of sexual selection in the long-tailed
manakin. Behavioral Ecology 4: 297-309.
Moss, R. 1987. Demography of capercaillie Tetrao urogallus in northeast Scotland. II.
Age and sex distribution. Ornis Scandinavica 18: 135-140.
35
Moyles, D. L., and D. A. Boag. 1981. Where, when, and how male sharp-tailed grouse
establish territories on arenas. Canadian Journal of Zoology 59: 1576-1581.
Pedersen, H. C. 1984. Territory size, mating status, and individual survival of males in a
fluctuating populatins of willow ptarmigan. Ornis Scandinavica 15: 197-203.
Riley, T. Z., C. A. Davis, M. Oritz, and M. J. Wisdom. 1992. Vegetative characteristics
of successful and unsuccessful nests of lesser prairie-chicken. Journal of Wildlife
Management 56: 383-387.
Robel, R. J. 1970. Possible role of behavior in regulating greater prairie-chickens. Journal
of Wildlife Management 34: 306-312.
Robel, R. J., J. N. Briggs, J. J. Cebula, N. J. Silvy, C. E. Viers, and P. G. Watt. 1970.
Greater prairie-chicken ranges, movements, and habitat usage in Kansas. Journal
of Wildlife Management 34: 286-306.
Robel, R. J., F. R. Henderson, and W. J. Jackson. 1972. Some sharp-tailed grouse
population statistics from South Dakota. Journal of Wildlife Management 36: 87-
98.
Sæther, B. E. 1990. Age-specific variation in reproductive performance of birds. Current
Ornithology 7: 251-283.
Sandercock, B. K., and A. Jaramillo. 2002. Annual survival of wintering sparrows:
assessing demographic consequences of migration. Auk 119: 149-165.
Schroeder, M. A. 1997. Productivity and habitat use of sharp-tailed grouse in north-
central Washington. Washington Department of Fish and Wildlife. Progress
Report. Olympia, Washington, USA.
Schroeder, M. A., and C. E. Braun. 1991. Walk-in funnel traps for capturing greater
prairie-chickens on leks. Journal of Field Ornithology 62: 378-385.
Storch, I. 2001. Capercaillie-BWP update. The Journal of birds of the Western Palearctic
36
3: 1-24.
Swenson, J. E. 1991. Social organization of hazel grouse and ecological factors
influencing it. Dissertation, University of Alberta, Edmonton, Alberta, Canada.
Szuba, K. J. and J. F. Bendell. 1988. Nonterritorial males in populations of spruce
grouse. Condor 90: 492-496.
U.S. Department of the Interior, Fish and Wildlife Service. 2002. Endangered and
threatened wildlife and plants; review of species that are candidates that are
proposed for listing as as endangered or threatened: the lesser prairie-chicken .
Fed. Regist. 67: 40667.
Waddell, B. and B. Hanzlick. 1978. The vanishing sandsage prairie. Kansas Fish and
Game Magazine1-4.
Weeden, R. B. 1965. Breeding density, reproductive success, and mortality of rock
ptarmigan at Eagle Creek, central Alaska from 1960 to 1964. Transactions of the
North American Wildlife and Natural Resources Conference 13: 336-348.
White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from
populations of marked animals. Bird Study 46: 1-31.
Wiley, R. H. 1974. Evolution of social organization and life history patterns among
grouse. Quarterly Review of Biology 49: 201-227.
Wittenberger, J. F. 1978. The evolution of mating systems in grouse. Condor 80: 126-
137.
Woodward, A. J., S. D. Fuhlendorf, D. M. Leslie, and J. Shackford. 2001. Influence of
landscape composition and change on lesser prairie-chicken (Tympanuchus
pallidinctus) populations. American Midland Naturalist 145: 261-274.
Zablan, M. A., C. E. Braun, and G. C. White. 2003. Estimation of greater sage-grouse
survival in North Park, Colorado. Journal of Wildlife Management 67: 144-154.
37
Tabl
e 1.
Num
bers
of a
dult
and
year
ling
less
er p
rairi
e-ch
icke
n m
ales
cap
ture
d fo
r the
firs
t tim
e (F
) or
reca
ptur
ed (R
) fro
m a
pre
viou
s yea
r, an
d tra
ppin
g ef
forts
in F
inne
y C
ount
y, K
ansa
s.
N
o. a
dults
No.
yea
rling
s
Tota
ls
Yea
r N
o. le
ks
Trap
day
sa D
ays/
lek
(SD
)
FR
FR
FR
1998
6
6811
.3 (1
.5)
44N
Db
50N
D94
ND
1999
862
7.8
(4.5
)26
623
1549
21
2000
c 20
138
6.9
(1.6
)85
552
1113
716
2001
1913
77.
2 (0
.5)
5945
4836
107
81
2002
1914
37.
5 (1
.2)
3826
3831
7657
Ove
rall
548
7.9
(1.7
)25
282
211
9346
317
5
38
a The
num
ber o
f day
s a se
t of w
alk-
in fu
nnel
s was
pla
ced
on a
lek.
b N
D =
no
data
.
c Tra
ppin
g w
as c
ondu
cted
on
Are
as I
and
II d
urin
g 20
00 to
200
2.
Table 2. Age-specific mark-recapture modeling for male lesser prairie-chickens in Finney
County, Kansas, 1998-2002.
Model statisticsa
Model structureb
AICc ∆AICc wi Dev K
φ 1age+t, φ2c, pc 653.92 0.00 0.749 22.93 7
φ t, pc 657.70 3.77 0.113 30.82 5
φ1t, φ2
c, pc 658.63 4.71 0.071 29.71 6
φ 1age*t, φ 2c, pc 660.09 6.17 0.034 22.85 10
φ 1age*t, φ 2t, pc 661.03 7.10 0.022 19.58 12
φ 1age*t, φ 2t, pt 662.67 8.75 0.009 16.97 14
φ 1age, φ 2c, pc 670.54 16.62 0.000 45.72 4
φ 1c, φ 2c, pc 672.12 18.20 0.000 49.33 3
φc, pc 678.41 24.48 0.000 57.64 2
a Model fit is described with deviance (Dev), the number of parameters (K), Akaike’s Information Criterion
corrected for small sample size (AICC), and AIC weights (wi). The candidate models are presented
including the global model, φ 1age*t, φ2t, pt.
b Models were structured to separate local survival for yearlings and adults (age) immediately after banding
(φ1), local survival in later transitions (φ2), and recapture rates (p). Factorial models (*) included main
effects and interaction with time; additive models (+) included main effects only. Time dependence (t) in a
given model is annual variation in rates.
39
Tabl
e 3.
Loc
al su
rviv
al ra
tes a
nd re
capt
ure
prob
abili
ties f
or m
ale
less
er p
rairi
e-ch
icke
ns se
para
ting
surv
ival
imm
edia
tely
af
ter f
irst c
aptu
re fr
om la
ter t
rans
ition
s, Fi
nney
Cou
nty,
Kan
sas,
1998
-200
2.
φ̂
afte
r 1st c
aptu
re ±
SE
φ̂
in la
ter y
ears
± S
E
cp̂±
SE
Yea
r
Yea
rling
A
dult
A
dult
Bot
h
1998
-99a
0.50
8 ±
0.08
9
0.32
0 ±
0.07
9
ND
b
0.74
0 ±
0.06
6
1999
-00
0.39
0 ±
0.09
7
0.22
6 ±
0.07
1
0.27
2 ±
0.09
4
0.74
0 ±
0.06
6
2000
-01
0.81
0 ±
0.07
1
0.66
1 ±
0.08
1
0.56
4 ±
0.14
3
0.74
0 ±
0.06
6
2001
-02
0.59
3 ±
0.09
5
0.40
0 ±
0.08
1
0.31
2 ±
0.06
0
0.74
0 ±
0.06
6
Ove
rallc
0.61
5 ±
0.06
8
0.48
5 ±
0.05
8
0.34
7 ±
0.04
7
0.73
4 ±
0.06
7
40
a Ann
ual v
aria
tion
of lo
cal s
urvi
val (
) est
imat
es w
ere
deriv
ed fr
om th
e m
odel
φ
φ̂1 ag
e *
t, φ
2 t, p c
.
b N
D =
no
data
. c O
vera
ll m
eans
wer
e de
rived
from
the
mod
el φ
1 age, φ
2 c, p
c.
Tabl
e 4.
The
num
bers
of w
ithin
- and
bet
wee
n-ye
ar in
terle
k m
ovem
ents
of b
ande
d m
ale
less
er p
rairi
e-ch
icke
ns, F
inne
y C
ount
y,
Kan
sas,
1998
-200
2.
B
etw
een
Yea
rs
W
ithin
yea
rs
Cap
ture
inte
rval
M
ovem
enta
Y
earli
ngA
dult
Yea
rling
Adu
lt
1-2
N
o48
(0.8
0)b
60 (0
.92)
34 (0
.83)
41 (0
.82)
1-2
Yes
12
(0.2
0)
5
(0.0
8)
7
(0.1
7)
9
(0.1
8)
2-3
No
9 (0
.69)
11 (1
.0)
23
(0.9
2)
14
(0.9
4)
2-3
Yes
Tota
ls
4 (0
.31)
03
(0.0
8)1
(0.0
6)
3-4
No
10
07
3-4
Yes
10
01
4-5
No
00
32
4-5
Yes
00
01
No
58 (0
.77)
71(0
.93)
60 (0
.86)
66 (0
.85)
Y
es
17 (0
.23)
5 (0
.07)
10 (0
.14)
12 (0
.15)
P-va
luec
0.00
40.
851
c The
P-v
alue
was
est
imat
ed u
sing
a li
kelih
ood
ratio
(G) t
est o
n a
2 ×
2 co
ntin
genc
y ta
ble
(yea
rling
/adu
lt: m
ove/
not-m
ove)
, df =
1, b
etw
een
year
s G=
8.22
2, a
nd w
ithin
yea
rs G
= 0
.035
.
b Num
ber a
nd p
ropo
rtion
(in
pare
nthe
ses)
of i
ndiv
idua
ls re
capt
ured
on
leks
with
in a
n ag
e-cl
ass (
colu
mn)
.
a Mov
emen
ts a
re d
escr
ibed
as a
bird
reca
ptur
ed a
t lea
st o
nce
on a
lek
othe
r tha
n th
e fir
st le
k of
cap
ture
.
41
Table 5. Estimates of survival for North American prairie grouse (Tympanuchus spp.) obtained from
live encounters (B), dead recovery (R), known-fate telemetry (T), and age-ratios from wing (W)
analyses.
Species
Mark type
Model typea
Sexb Survival Reference
LPCHc B CJS M 0.46 This Study
LPCH R LFT M 0.32 Campbell (1972)
LPCH T K-M M 0.57 Jamison (2000)
GPCHc B LFT M 0.48 Hamerstrom and Hamerstrom (1973)
SHTGc B LFT M 0.50 Amman (1957)
SHTG R LFT U 0.25 Robel et al. (1972)
SHTG B PRP M 0.17 Moyles and Boag (1981)
SHTG B PRP M 0.36 Gratson et al. (1991)
SHTG T K-M U 0.57 Schroeder (1997)
SHTG W PRP U 0.56 Giesen (1999)
SHTG T K-M M 0.35 Boisvert (2000)
a Model types are as follows: CJS = Cormack-Jolly-Seber, K-M= Kaplan-Meier, LFT = life-table
analysis, and PRP = the proportion of survivors reported.
b Studies that were male specific Sex = M, and in those that did not differentiate gender in estimates
Sex = U.
c LPCH = T. pallidicinctus, GPCH = T. cupido, SHTG = T. phasianellus
42
Table 6. Estimates for age-specific survival and breeding for male grouse obtained from live encounters
(B), dead recovery (R), known-fate telemetry (T), and age-ratios from wing (W) analyses.
Age / survival rate Delayed maturation
Mating system / species
Model typea
Data 1b 2 ≥2 Refc Y Refc
Monogamous
Willow ptarmigan PRP B 0.44 NDd 0.25 5 N 5,17
Rock ptarmigan PRP B 0.24 ND 0.21 1 N 1
White-tailed ptarmigan LFT B 0.76 0.74 0.30 3 N 3
Hazel grouse K-M T 0.75 ND 0.75 21 N 18
Dispersed arena Ruffed grouse PRP B 0.41 0.42 0.44 4 Y 4
Siberian spruce grouse ND ND ND ND ND Y 19
Spruce grouse PRP B 0.92 ND 0.81 9 Y 15
Blue grouse PRP B 0.55 0.84 0.75 11 Y 11
Lek mating Greater prairie-chicken LFT B 0.50 0.48 0.41 7 Y 6,7
Lesser prairie-chicken CJS B 0.62 0.49 0.35 23 Y 23
Sharp-tailed grouse PRP B 0.39 ND 0.33 16 Y 10, 16
Black grousee K-M T 1.0 ND 0.70 9 Y 2
Greater sage-grouse REC B 0.63 ND 0.37 22 Y 8,14
Capercaillie PRP W 0.54 ND 0.46 13 Y 20
a Model types are as follows: CJS = Cormack-Jolly-Seber, K-M= Kaplan-Meier, LFT = life-table
analysis, and PRP = the proportion of survivors reported.
43
Table 6. continued b Age of birds 1 = yearling, 2 = 2 years old (adult in this study), and ≥ 2 = 2 years old or older (older
adult in this study). c References: 1, Weeden (1965); 2, Kruijt and Hogan (1967); 3, Braun (1969) 4, Gullion and Marshall
(1968); 5, Bergerud (1970); 6, Robel (1970); 7, Hamerstrom and Hamerstrom (1973); 8, Wiley (1974);
9, Keppie (1979); 10, Moyles and Boag (1981); 11, Lewis and Zwickel (1982); 12, Anglestam (1984);
13, Moss (1987); 14, Hartzler and Jenni (1988); 15, Szuba and Bendell (1988); 16, Gratson et al.
(1991); 17, Martin (1991); 18, Swenson (1991); 19, Andereev et al. (2001); 20, Storch (2001); 21,
Montadert and Leonard (2002); 22, Zablan et al. (2003); 23, This study. d ND = no data available. e Survival estimates were for breeding and summer season only.
44
CHAPTER 2
RADIOTELEMETRY ESTIMATES OF SURVIVAL IN THE LESSER
PRAIRIE-CHICKEN: ARE THERE TRANSMITTER BIASES?
Abstract: Radiotelemetry has provided wildlife biologists with a tool to estimate survival
where fate of each individual is likely known. Whereas analyses of these data can result
in highly accurate estimates, 5 assumptions must be met. Two of these assumptions, that
right-censoring is random with respect to the survival of study animal, and that
transmitters have no effect on survival of study animal, are often difficult to assess.
Using joint models developed for live encounter and dead recovery data, I examined the
potential for bias in survival estimates of radiomarked male lesser prairie-chickens in
southwestern Kansas. Additionally, I use graphing techniques to assess if the assumption
of random censoring holds in this study. In total, 201 male lesser prairie-chickens were
captured and marked during this study. Seventy-five of these birds were fitted with
radios (68 survived a 2-week acclimation period) and 126 were in the banded group.
Model selection and parameter estimation were based on the information theoretic-
approach. The model best supported by the data, Sc, pgroup+t, rg, Fc, indicated that survival
was best modeled as constant (Ŝc = 0.731, SE = 0.072) across radiomarked and banded
birds. Signal loss occurred throughout the monitoring period and appeared to be
independent of periods of high mortality. Eight of 16 (50.0, SE = 1.6 %) right-censored
birds were subsequently recaptured, which was similar to the recapture rates for known-
fate birds (23 of 52; 55.8%, SE = 0.5 %), indicating that right-censored birds had similar
survival rates to that of known-fate individuals.
45
INTRODUCTION
Radiotelemetry is widely used to collect data for estimating survival rates of
experimental treatments or groups in wildlife studies. Unbiased survival estimation
requires meeting several fundamental assumptions, 1) radioed animals are representative
(random and independent samples) of the population of interest, 2) observation periods
(or locations) are independent, 3) working radios are always located, 4) censoring is
random (i.e., any animal not located is as likely to be alive as dead), and 5) radios do not
impact survival of marked individuals (Winterstein et al. 2001). Assumptions 1 to 3 can
be met by implementing the appropriate research design. Assumptions 4 and 5 are less
controllable as the vagaries of equipment and behavior of individual animals cannot be
accounted for entirely in project design. However, assumptions 3 and 4 are closely
related as undetected individuals (i.e., vagaries of equipment) will be right-censored for a
given period. Right-censoring occurs when an individual “leaves” a study either
temporarily or permanently and the fate of the individual is unknown. Thus, all
individuals are “right-censored” at the termination of a project because, by definition,
their fate beyond that point is unknown. In practice animals are right-censored when
there is radio failure, detection rates are < 1, or there is emigration from the study area.
Often the detection of a working radio depends on the length of the observation period,
and proper planning of a project can increase the likelihood of meeting assumption 3.
Generally, researchers are faced with accepting assumptions 4 and 5 without controls to
test their validity or impact (Esler et al. 2000).
The effect of radio transmitters on survival on game birds has been well-studied,
but the results have been mixed. Most studies suffer from a lack of a suitable control
46
group and rigorous estimates of survival. Typically studies examine differences in return
rates or daily survival of birds marked with 2 or more types of harness configuration or
transmitter mass. Several studies on upland game birds provided evidence suggesting
that radios with external attachments (e.g., poncho, necklace, and backpack) had negative
impacts on survival (see review by Withey et al. 2001). In contrast, other study designs
have found no measurable effect of radios on survival or other vital rates (Hines and
Zwickel 1985, Cotter and Gratto 1995, Thirgood et al. 1995, Bro et al. 1999). Handling
effects (i.e., increased mortality immediately after handling) have occurred with early
radio designs (Marks and Marks 1987, Pekins 1988). However, the studies mentioned
above examined return rates as a measure of survival. Unfortunately return rates are
confounded by 4 probabilities: true survival (S), site fidelity (F), site propensity (γ*), and
detection (p*). Mark-recapture models can provide survival estimates while accounting
for fidelity and detection rates. However, return rates can be biologically meaningful
when a species exhibits fidelity to trapping areas (Esler et al. 2000). Currently, no studies
have examined the effects of radiomarking using contemporary analyses of capture-
recapture.
I examined the potential effect of radiomarking on survival of male lesser prairie-
chicken (Tympanuchus pallidicinctus) using a novel approach of live-recapture dead
recovery models. Typically, joint-analyses of live and dead recoveries (Burnham 1993)
have been applied to banded individuals marked-recaptured on breeding grounds and
possibly recovered during the hunting season. I expanded these joint-analyses to include
live capture and recapture of banded and radiomarked (i.e., physical recapture or
detection of transmitter signal from live radiomarked birds) individuals, and dead
47
recovery of hunter-harvested (both banded and radiomarked) and mortalities of
radiomarked individuals. I also examined the potential bias of right-censoring by
comparing the return rates of known-fate and right-censored individuals.
METHODS
Trapping, marking, monitoring
This study was conducted in native sand sagebrush prairie south of Garden City,
Kansas (37° 52′ N, 100° 59′ W) from spring 1998 to spring 2000. Lesser prairie-
chickens were captured over 3-week periods using walk-in funnel traps on leks during
spring (in late Mar to Apr) and fall (late Sep to Oct) (Haukos et al. 1991, Salter and Robel
1999). Both females and males were captured but recaptures of females were few and
this study examines only the male cohort of the radiomarked population. Captured birds
were marked with serially numbered aluminum leg bands. Relatively equal numbers of
birds were marked at all leks (n = 11) to ensure relatively equal sampling across the study
site. Birds were fitted with necklace-style radios with a mass ≤ 12 g; ≤ 1.7% of a male’s
body mass ( x = 790 g, range = 700 - 950 g). The radio transmitters had 8-month battery-
life, whip antennae, and 8 hr mortality switch. All birds that did not survive 2-weeks
post-capture (i.e., acclimation period) were included in the analysis. This conservative
approach allowed for acute effects of transmitters to be considered. Radiomarked birds
were monitored daily using a truck-mounted null-peak antenna system to ascertain their
status. For purposes of this study, the observation period was 6 months for radiomarked
birds.
48
Data analysis
I used a joint analysis for mark-recapture and staggered-entry known-fate data
from radiotelemetered and banded birds using live-recapture dead-recovery models
(hereafter joint models) (Burnham 1993) to examine the effects of radiomarking on lesser
prairie-chicken survival. My analysis included 5 encounter occasions (5 6-month
intervals between April 1998 and April 2000) and 2 attribute groups (banded or
radiomarked). Radiomarked individuals that survived a given interval and had their radio
removed at recapture (n = 11) were right-censored from the radiomarked group and
added to the banded group, and the opposite was true for banded birds that received a
radio at a subsequent interval (n = 5). When a radio signal was lost prior to a capture
period (i.e., prior to the expected life of the transmitter) they were treated as a removal
from the encounter history and right-censored.
Joint models of survival allow for combination of multiple sources of information
(Burnham 1993, Barker and White 2001). In my study, these sources of live encounter
come from 5 standard mark-recapture periods (once every 6 months) of banded birds,
“resighting” (p) periods of live radiomarked individuals, and reporting periods (r) of both
banded and radiomarked birds. Although hunter reporting of banded birds did occur, this
proportion was < 5%, but this information was included in the analysis. Thus, a
radiomarked individual may be alive and encountered, but not physically recaptured at a
given encounter period. Joint models necessarily are more complex than Cormack-Jolly-
Seber (CJS) models of live capture data (Burnham 1993). The parameters and their
definitions are as follows:
49
Si = The probability of survival; i.e., an animal that is alive at time i, is again alive at time
i + 1.
pi = The probability of detection; i.e., an animal that is alive, is at risk of capture at
trapping occasion i is captured.
ri = The probability of reporting; i.e., a marked animal that is alive at time i dies is
reported at time i + 1.
Fi = The probability of site fidelity; i.e., an animal that is at risk of capture at trapping
occasion i and is again at risk of capture at time i + 1.
Joint models of survival of banded and radiomarked individuals were analyzed
following three steps: 1) selection of the global model, 2) goodness-of-fit tests (GOF),
and 3) development of less parameterized models and model selection. Model selection
criterion was based on either Akaike’s Information Criterion adjusted for small sample
sizes (AICc), or the quasi-likelihood corrected version of AICc (QAICc). Protocols of
Burnham and Anderson (1998) were followed to identify the model best supported by the
data. Model development and parameter estimation was conducted in MARK 3.0 (White
and Burnham 1999) using design matrices and the logit link function.
I developed an a priori set of candidate models (Table 1) that allowed me to
examine the potential for the effects of radiomarking. My primary interest was
evaluating survival (S); thus, recapture (p), reporting (r), and fidelity (F) were considered
nuisance parameters. Subscripts following parameters indicate explanatory variables
(e.g., S group*t describes a model that includes both marker effect (i.e., group = marker type
banded or radiomarked) and time-dependence (i.e., survival varies by encounter
occasion) in survival. Subscript (handle) denotes different survival rate for the interval
50
after initial release or all subsequent intervals, respectively. Lastly, I used (c) to describe
models with parameters held constant (e.g., Sc, describes a model where survival is held
constant). I fitted alternative structures of F, r, and p to the global model (S group*t, pgroup+t,
rgroup, Fc) that were biologically and practically meaningful. I fixed the parameter Fc =
0.979 as only 2 of 95 (0.979) radiomarked males were known to have permanently
emigrated. Because few banded birds were reported dead, I did not consider time-
dependence in r. I assumed that behavioral differences in seasons would lead to
consistently unequal capture probabilities, p, between fall and spring encounters (Salter
and Robel 1999), and additive time structure was assumed for p under the global model.
Subsequently, time (t) and group (group) structures were fitted to p and the best structure
was selected based initially on minimum AICc. Given that the detection probabilities in p
and r of radiotelemetry (~1) and banding data (< 1) were unequal, I identified a limited
number of structures for these parameters. Specifically I did not consider models with p
and r held constant. The simplest form was with a group effect only. Survival of
radiomarked and banded birds was examined as a group effect (Sgroup). An alternative
model structure was fit to examine potential differences in handling effect for both
marker types (Sgroup, handle). This model allowed survival to vary by group effect, and for
the interval of capture to be different from later encounters.
The overall GOF of the global model was assessed using the parametric bootstrap
in MARK. The proportion of simulations (n = 1,000) that had a greater deviance than the
observed deviance provided an assessment of model fit. If this proportion was small
(<0.05) then this provides evidence that the model fits the data poorly (Cooch and White
2001). I used these simulations to estimate the degree of overdispersion (ĉ), by dividing
51
the observed deviance by the bootstrapped mean deviance. If the global model had a
poor fit, and ĉ < 3, then AICc was adjusted accordingly to this derived value of ĉ, and
QAICc used for model selection (Anderson et al. 1994).
I examined the assumption of random censoring by comparing return rates of
known-fate individuals (number alive and recaptured / number alive + number dead)
were compared to the proportion of birds that initially were right-censored, but were
recaptured or reported later (number censored and recaptured / total number censored). I
constructed a 95% confidence interval around the difference of these proportions to
determine if it was different from 0 (Agresti 1996).
I calculated an effect size (i.e., difference of means) and its 95% confidence
interval to determine if body masses differed between the 2 groups. I report sample sizes
and standard deviation along with the effect size.
RESULTS
A total of 201 male lesser prairie-chickens was captured and marked during this
study. Seventy-five of these birds were fitted with radios (68 survived the 2-week
acclimation period), and 126 were in the banded treatment group. Fall trapping resulted
in fewer birds captured per unit effort than spring trapping (Salter and Robel 1999), and
this was evident in the group and time additive (group + t) parameter structures in the
model selection below. Treatment groups were similar with respect to body mass and
age-classes. Body mass of radiomarked (n = 72, x = 790, SD = 48 g) and banded (n =
133, x = 803, SD = 44 g) birds was similar ( x difference = −12.2, CI: −25.5, 1.1 g) as
was the proportion of yearlings to adults in each sample, 0.83 and 0.79, respectively.
52
The fit of the global model (Sgroup*t, pgroup+t, rg, Fc) was rejected by the parametric
boostrap (P < 0.001), but was retained because overdispersion was minor ĉ = 2.382.
Model selection and inference were made from QAICc (Anderson et al. 1994). Some of
the overdispersion may have been due to a lack of independence, as some individuals
were in both the banded and radiomarked groups.
The model best supported by the data, Sc, pgroup+t, rg, Fc, indicated that survival
was best modeled as constant (Ŝc = 0.742, SE = 0.065) across radiomarked and banded
birds. There was some model uncertainty as ∆QAICc < 2 for the top 3 models, which
suggested that handling, S handle, pgroup+t, rg, Fc (Ŝ1 = 0.704, SE = 0.086; Ŝ2 = 0.797, SE =
0.114), and group effects, Sgroup pgroup+t, rg, Fc (Ŝradio = 0.761, SE = 0.069; Ŝband = 0.671,
SE = 0.112) may have been present (Table 1). The survival estimates from model
averaging indicated that radiomarked individuals had slightly higher survival (Ŝradio =
0.745, SE = 0.073) than banded birds (Ŝband = 0.711, SE = 0.090). Reporting rates were
higher for radiomarked birds (r = 0.636, SE = 0.157) than banded birds (r = 0.103, SE =
0.057). Recapture probabilities varied additively over time for both groups (Table 2),
and, radiomarked birds had higher overall recapture rates (pradio = 0.645, SE = 0.120;
pband = 0.303, SE = 0.124).
Effects of right-censoring
Radio signals were permanently lost for 16 birds during the 6-month sampling
intervals and were right-censored. An additional 24 radio signals were lost after the 6-
month battery life expired, and were detected as alive at last encounter occasion (this
accounts for part of known-fate percentages). Signal loss occurred throughout the
monitoring period and appeared to be independent of periods of high mortality (Fig. 1).
53
Eight of 16 (50.0, SE = 1.6 %) right-censored birds were subsequently recaptured, which
was similar (% difference = −5.8; CI: −33.8, 22.2 %) to the recapture rates for known-
fate birds (23 of 52; 55.8%, SE = 0.5 %), suggesting that right-censored birds had similar
survival rates to that of known-fate individuals.
DISCUSSION
This study found no evidence that radiomarking male lesser prairie-chickens
negatively impacted their survival. Rather, contrary to the predicted outcome,
radiomarked individuals had a slightly higher survival rate. Right-censored birds had
similar return rates when compared to those with known-fates. Inference about survival
in this group was based on relatively small sample sizes but return rates were similar
enough to provide support for a comparable mortality rate between them. I recognize
that my analyses were retrospective and an experimental design would have had greater
statistical power. However, given the positive direction of the difference between
radiomarked and banded birds, it is likely that radios had little impact on bird survival in
my study population. Inference in this study is limited to the male cohort of the
population. However, given that females average about < 70 g males, it is likely that
small transmitters (<12 g) have little impact on their survival.
Two types of transmitters
This is the first study to demonstrate that radios had no measurable effect on
prairie grouse (Tympanuchus spp.). Previous work on both sharp-tailed grouse (T.
phasianellus) and greater prairie-chickens (T. cupido) detected large negative effects of
radios on survival (Marks and Marks 1987, Burger et al. 1991). It is important to note
that Marks and Marks (1987) and Burger et al. (1991) measured the effects of the larger
54
(18-22 g) poncho-style solar powered transmitters, in contrast to the 12-g lithium battery-
powered necklace style used in my study. Marks and Marks (1987) lacked a suitable
control group (n = 9) to compare return rates or survival with the radiomarked birds (n =
36). However, 23 of 35 radiomarked birds in the Marks and Marks (1987) study
succumbed to predation, of which 65% occurred during the breeding season. Burger et
al. (1991) documented that birds marked with larger 2-stage transmitter and reflective
surface had twice the probability of being depredated than those fitted with 1-stage radios
(14 g). Similar patterns of mortality were documented for ruffed grouse (Bonasa
umbellus) and ring-necked pheasant (Phasianus colchicus), where birds marked with a
heavier backpack style transmitters had reduced survival rates (Warner and Etter 1983,
Small and Rusch 1985).
Two types of marking
Thirgood et al. (1995) found no measurable effect of radiomarking (necklace style
transmitters) on red grouse (Lagopus lagopus scoticus) when compared to patagium
tagged birds. They suggested that the miniaturization of the necklace-style transmitters
has alleviated some of the lethal effects reported earlier in other grouse species with
larger transmitters (Marks and Marks 1987, Pekins 1988). Thirgood et al. (1995)
suggested that necklace style transmitters are the preferred method for marking grouse
species. Alternatively, Caizergues and Ellison (1998) reported lower nest success rates
and smaller brood sizes of radiomarked black grouse (Tetrao tetrix) females when
compared to unmarked individuals, but they did not examine survival rates. Overall,
radiomarked females raised fewer broods than unmarked birds, but chicks per female
were nearly equal between the 2 groups (Caizergues and Ellison 1998). One limitation to
55
their study was that individuals were not individually identifiable in their unmarked
population. Thus, larger brood sizes could have been an artifact of double counting
females with large broods. The marked sample was relatively small (≤ 8 marked females
per year, n = 26 for 5 years), thus detecting annual differences in reproductive success
between groups was not possible.
The cause of marker-specific survival rates may not be easily understood as
yearly variation can impact the outcome of these studies (Cotter and Gratto 1995, Bro et
al. 1999). Years when birds may be stressed or in poor condition may increase the
negative effect of radiomarking. In my study, models with time dependency (i.e., either
Sgroup+t or Sgroup*t) in survival were not supported by the data. Although the additive
models had slightly better fit than full time-dependence, they indicated that radioed birds
on average had higher survival than banded birds.
Censoring
Random censoring is a critical assumption of telemetry studies, and violation of
the assumption can result in significant bias in Ŝ (Tsai et al. 1999, Winterstein et al.
2001). This assumption has been difficult to evaluate in the field (Esler et al. 2000). In
this study, return rates were similar between known-fate and right-censored individuals,
and the timing of censoring appeared to be independent of mortality (Fig. 1). However,
the power of my retrospective study to detect even a 10 % difference in return rates was
low (β = 0.11). Esler et al. (2000) documented a similar pattern in radiomarked harlequin
ducks (Histronicus histronicus) but also acknowledged the low power of their study
design.
56
Miniaturization and method of attaching transmitters appear to have reduced the
impact of radios on survival of prairie grouse, because survival rates of radioed birds in
my study were higher than those reported by Marks and Marks (1987) and Burger et al.
(1991). Male lesser prairie-chickens in my study were not measurably impacted by
radiomarking, as their survival rates were greater than or equal to those of banded birds.
Inference was limited to the male cohort of this population. Future work should
implement a rigorous experimental study to examine the effects of transmitters on free-
living birds and females under field conditions.
ACKNOWLEDGMENTS
I thank J. O. Cattle Co., Sunflower Electric Corp., Brookover Cattle Co., and P. E.
Beach for access to private property. B. E. Jamison, J. C. Pitman, G. C. Salter, T. G.
Shane, and T. L. Walker, Jr., assisted with trapping and banding of prairie chickens. B.
K. Sandercock and G. C. White provided useful insights to the analysis of the joint live-
dead recovery models. Financial and logistical support was provided by Kansas
Department of Wildlife and Parks (Federal Aid in Wildlife restoration projects, W-47-R
and W-53-R), Kansas Agricultural Experiment Station (Contribution No. 03-_______),
Division of Biology at Kansas State University, and Weststar Energy Inc.
LITERATURE CITED
Agresti, A. 1996. An introduction to categorical data analysis. Wiley & Sons Inc.. New
York, New York, USA.
Anderson, D. R., K. P. Burnham, and G. C. White. 1994. AIC model selection in
overdispersed capture-recapture data. Ecology 75: 1780-1793.
Barker, R. J., and G. C. White. 2001. Joint analysis of live and dead encounters of
57
marked animals. Pages 361-367 in R. Field, R. J. Warren, H. Okarma, and P. R.
Sievert, editors. Wildlife, land, and people: priorities for the 21st century.
Proceedings from the Second International Wildlife Congress. The Wildlife
Society, Bethesda, Maryland, USA.
Bro, E., J. Clobert, and F. Reitz. 1999. Effects of radiotransmitters on survival and
reproductive success of gray partridge. Journal of Wildlife Management 63: 1044-
1051.
Burger, L. W., M. R. Ryan, D. P. Jones, and A. P. Wywialowski. 1991. Radio
transmitters bias estimation of movements and survival. Journal of Wildlife
Management 55: 693-697.
Burnham, K. P. 1993. A theory for combined analysis of ring recovery and recapture
data. Pages 199-213 in P. North and J. D. Lebreton, editors. Marked individuals
in bird population studies. Birkhauser Verlag, Basel, Switzerland.
Burnham, K. P., and D. R. Anderson . 1998. Model selection and inference a practical
information-theoretic approach. Springer. New York, New York, USA.
Caizerigues, A., and L. N. Ellison. 1998. Impact of radio-tracking on black grouse
Tetrao tetrix reproductive success in the French Alps. Wildlife Biology 4: 205-
212.
Carroll, J. P. 1990. Winter and spring survival of radio-tagged gray partridge in North
Dakota. Journal of Wildlife Management 54: 657-662.
Cooch, E., and G. C. White 2001. Using MARK–a gentle introduction.
http://www.cnr.colostate.edu/~gwhite/mark/mark.html (6 June 2002)
Cotter, R. C., and C. J. Gratto. 1995. Effects of nest and brood visits and radio
transmitters on rock ptarmigan. Journal of Wildlife Management 59: 93-98.
Esler, D., D. M. Mulcahy, and R. L. Jarvis. 2000. Testing assumptions for unbiased
estimation of survival of radiomarked harlequin ducks. Journal of Wildlife
58
Management 64: 591-598.
Haukos, D. A., L. M. Smith, and G. S. Broda. 1990. Spring trapping of lesser prairie-
chickens. Journal of Field Ornithology 61: 20-25.
Hines, J. E., and F. C. Zwickel. 1985. Influence of radio packages on young blue grouse.
Journal of Wildlife Management 49: 1050-1054.
Jamison, B. E., R. J. Robel, J. S. Pontius, and R. D. Applegate. 2002. Invertebrate
biomass: associations with lesser-prairie chicken habitat use and sand sagebrush
density in southwestern Kansas. Wildlife Society Bulletin 30: 517-526.
Marks, J. S., and V. S. Marks. 1987. Influence of radio collars on survival of sharp-tailed
grouse. Journal of Wildlife Management 51: 468-471.
Pekins, P. J. 1988. Effects of poncho mounted radios on blue grouse. Journal of Field
Ornithology 59: 46-50.
Salter, G. C., and R. J. Robel. 2000. Capturing lesser prairie-chickens on leks during fall.
Transactions of the Kansas Academy of Science 103: 46-47.
Small, R. J., and D. H. Rusch. 1985. Backpacks vs. ponchos: survival and movements of
radio-marked ruffed grouse. Wildlife Society Bulletin 13: 163-165.
Thirgood, S. J., S. M. Redpath, P. J. Hudson, M. M. Nichols, and N. J. Aebischer. 1995.
Effects of necklace radio-transmitters on survival and breeding success of red
grouse Lagopus lagopus scoticus. Wildlife Biology 1: 121-126.
Tsai, K., K. H. Pollock, and C. Brownie. 1999. Effects of violation of assumptions for
survival analysis methods in radiotelemetry studies. Journal of Wildlife
Management 63: 1369-1375.
Warner, R. E., and S. L. Etter. 1983. Reproduction and survival of radio-marked hen
ring-necked pheasants in Illinois. Journal of Wildlife Management 47: 369-375.
White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from
59
populations of marked animals. Bird Study 46: 1-31.
Winterstein, S. R., K. H. Pollock, and C. M. Bunk . 2001. Analysis of survival data from
telemetry studies. Pages 351-380 in J. Millspaugh and J. Marzluff, editors. Radio
Tracking and Animal Populations. Academic Press, San Diego, California, USA.
Withey, J. C., T. D. Bloxton, and J. M. Marzluff . 2001. Effects of tagging and location in
error in wildlife radiotelemetry studies. Pages 45-75 in J. Millspaugh and J.
Marzluff, editors. Radio Tracking and Animal Populations. Academic Press,
San Diego, California, USA.
60
Table 1. Candidate live-dead recovery models for estimating survival of radiomarked
and banded lesser prairie-chickens in Finney County, Kansas, 1998-2000.
Model statisticsa
Model structureb
QAICc ∆QAICc wi K Dev
Sc, pgroup+t, rgroup, Fc 279.5 0.00 0.43 8 77.81
Sc,handle, pgroup+t, rgroup, Fc 280.8 1.36 0.22 9 77.04
Sgroup, pgroup+t, rgroup, Fc 281.1 1.65 0.19 9 77.33
Sgroup,handle, pgroup+t, rgroup, Fc 282.5 3.02 0.09 10 76.55
St, pgroup+t, rgroup, Fc 284.9 5.42 0.03 12 74.62
Sgroup + t, pgroup+t, rgroup, Fc 286.2 6.72 0.01 13 73.73
Sgroup*t, handle, pgroup+t, rgroup, Fc 287.9 8.37 0.01 18 64.23
St,handle, pgroup+t, rgroup, Fc 288.1 8.57 0.01 13 75.58
Sgroup + t ,handle, pgroup+t, rgroup, Fc 288.2 8.70 0.01 14 73.51
Sgroup, handle+ t, pgroup+t, rgroup, Fc 288.6 9.15 0.00 14 73.96
Sgroup*t, pgroup+t, rgroup, Fc 288.8 9.30 0.00 15 71.89
Sgroup + t ,handle+t, pgroup+t, rgroup, Fc 292.2 12.72 0.00 17 70.84
Sgroup* t ,handle+t, pgroup+t, rgroup, Fc 293.3 13.82 0.00 22 60.48
a Model fit is described by deviance (Dev), the number of parameters (K), quasi-Akaike’s
Information Criteria (ĉ = 2.382) corrected for small sample size (QAICc), and QAICc
weights (wi).
b Models were structured to separate survival (Ŝ), for marker effect (group) for both
radiomarked and banded males immediately after handling (handle), and survival in later
transitions. Parameter estimates include recapture rate (p), reporting rate (r), and site
fidelity (F) which was fixed at 0.979.
61
Table 2. Model averaged parameter estimates (± SE) from the 3 best (∆QAICc< 2) live-
dead recovery models of radiomarked and banded male lesser prairie-chickens in Finney
County, Kansas, 1998-2000.
Marker type
Parameter estimatesa Banded Radio
S group 0.711 ± 0.090 0.745 ± 0.073
pfall-98 0.161 ± 0.076 0.502 ± 0.129
pspring-99 0.574 ± 0.162 0.876 ± 0.071
pfall-99 0.158 ± 0.084 0.495 ± 0.137
pspring-00 0.318 ± 0.172 0.707 ± 0.145
r group 0.103 ± 0.057 0.636 ± 0.157
F 0.979 ± 0.000 0.979 ± 0.000 aParameter estimates include survival (Ŝ), recapture rate (p), reporting rate (r), and site
fidelity (F).
62
Fig 1. The cumulative frequency distribution (A) of mortalities and right-censoring
indicates that the high mortality periods (2-week intervals) were not associated with right
censoring. The distribution of mortality and right-censoring (B) indicates the actual
numbers lost to mortality or signal in a given 2-week period were not usually related.
Both figures show time (2-week periods) from capture and release.
63
10 20 30 40
Cum
ulat
ive
Freq
uenc
y D
istri
butio
n
0.0
0.2
0.4
0.6
0.8
1.0
DeadCensored
Expected date of radio failure
2
2-week periods after initial capture
10 20 30 40
Num
ber o
f bird
s
0
2
4
6
8
dead censor
Expected date of radio failure
2
(A)
(B)
64
CHAPTER 3
GENDER AND AGE-SPECIFIC SURVIVAL AND PROBABLE CAUSES OF
MORTALITY IN RADIOMARKED LESSER PRAIRIE-CHICKENS IN KANSAS
Abstract: Long-term declines in population indices and reductions in habitat have
lead to increasing concern over the status of the lesser prairie-chicken (Tympanuchus
pallidicinctus). It is essential to identify management approaches that have the greatest
positive impact on rates of population growth. I evaluated the effects of season, age, and
gender on the survival of lesser prairie-chickens in southwestern Kansas. Using data
from a 6-year field study, I estimated summer (Apr - Nov) and seasonal rates (Apr - Mar)
using known-fate (radiotelemetry) survival models. I evaluated the probable mortality
causes of radiomarked birds. In total, 136 male (46 yearling, 66 adult, 24 age
undetermined), and 227 female (87 yearling, 117 adult, and 23 age undetermined) lesser
prairie-chickens were captured and fitted with radio transmitters from 1997 to 2002. The
summer known-fate analyses revealed that overall male (Ŝ = 0.71, SE = 0.06) and female
(Ŝ = 0.69, SE = 0.06) survival rates were similar, but females were most susceptible to
mortality during May and June. Additionally, the summer analysis indicated that
yearling females had a greater probability (Ŝ = 0.77, SE = 0.06) of surviving than adults
(Ŝ = 0.62, SE = 0.05). These patterns were consistent for both the seasonal and summer
survival rates. One-hundred and thirty-four mortalities were recorded from spring 1997
to spring 2003, and 54% were attributed to mammalian predation. Generally, lesser
prairie-chickens had higher estimated survival than those reported for other prairie
grouse, suggesting that adult survival was not a limiting factor for lesser prairie-chickens
in Kansas.
65
INTRODUCTION
Seasonal survival (S) rates and the timing of mortality (1 - S) events are important
parameters in evolutionary ecology and wildlife management (Caizergues and Ellison
1997). Understanding the seasonal variation in the timing of mortality and its severity
are especially important to grouse management (Bergerud 1988), as females in
promiscuous systems (unipaternal care) may have an increased mortality risk during
incubation and brood-rearing periods. Yearling and adult females may have different
survivorship based on reproductive effort and success (Saether 1990). In general, more
experienced birds will have greater fecundity and survival, thus resulting in lower
yearling survival rates (Saether 1990).
Male and female grouse of promiscuous mating systems have different
reproductive costs. Males do not participate in nest-building, incubation, or brood-
rearing, their only contribution is gametes. However, males do incur a cost of breeding
because of conspicuous plumage and displays during territorial defense and may suffer
higher mortality than females. Females incur a cost of reproduction through egg
production, incubation, and brood rearing. Their susceptibility to predation during
incubation is thought to be higher than other times of year (Bergerud 1988). It is unclear
if a short-term decrease in female survival (reproductive cost) would yield a lower annual
survival rate than males. Is it possible that the cost of nesting and conspicuous display
will result in similar survival rates between the genders?
Differential survival in gender and age-classes may result in either skewed sex- or
age-ratios in harvest (e.g., wing or head samples) or count data (Amman 1957, Campbell
1972, Linden 1981, Moss 1987, Bergerud 1988). Survival estimates from age-ratios
66
assume stable populations, equal sampling of yearlings and adults, and a constant
mortality rate. Most survival estimates for grouse have been based on these methods
(Bergerud 1988). Given the tenuous status of many grouse populations (Storch 2000),
more accurate estimates are needed for conservation actions. Radiotelemetry may
provide accurate parameter estimates if the basic assumptions are met (Winterstein et al.
2001).
Little is known about seasonal or annual survivorship of lesser prairie-chickens
(Tympanuchus pallidicinctus), but such information is critical for this species of
conservation concern. It is currently listed as a “warranted but precluded” threatened
species (U.S. Fish and Wildlife Service 2002). Band recovery data from hunters in New
Mexico were used to estimate an annual survival rate of 35% (Campbell 1972), and
Merchant (1982) estimated a 6-month survival rate of 59% (extrapolated 12-month
survival = 35%) for radiomarked females that included the nesting and brood rearing
periods. In Kansas, Jamison (2000) estimated annual survival of radiomarked males to
be 57%. Six-month (Apr-Sep) survival rates were estimated at 74% for males and
females during the same study (Jamison 2000). This estimate extrapolated to 12 months
would be approximately 55%. However, lower survival rates of females than males have
been implicated in biased sex-ratios (Campbell 1972); there is little evidence to support
this idea. The available information on lesser prairie-chicken survival is incomplete, as
there are few annual estimates, and no information on the timing of mortality for females
and males during a year. Such information is imperative for conservation efforts (e.g.,
population viability analyses, harvest regulations) and basic ecological understanding.
67
Using radiotelemetry data from 1997 to 2003, I address 3 research hypotheses and
provide a description of probable mortality causes of the lesser prairie-chicken. I
examined survival in terms of an overall rate (9 and 12-month periods for males and
females) to identify potential cumulative effects of gender-specific costs of reproduction.
I quantified the timing of mortality events for males and females to verify the costs of
reproduction hypothesis (i.e., did both males and females suffer the lowest survival rates
during the breeding season?). Thus, if the cost of reproduction hypothesis is supported
then timing may not differ but the overall survival rate will differ. I estimated survival
for yearling and adult females to determine if age-structured survival was evident in the
population. Here, I test the constraint hypothesis, which argues that yearlings lack some
of the skills essential for high reproductive performance. Because high reproductive
performance is often linked to survival (Saether 1990), I used survival and ancillary data
on reproductive rates to evaluate this hypothesis. Lastly, I evaluated the probable
mortality causes of radiomarked birds.
METHODS
Study area
The study region was comprised of 2 fragments (~5000 ha each) of native
sandsage (Artemisia filifolia) prairie near Garden City, Finney County, Kansas (37° 52′
N, 100° 59′ W). Work began on Area I (southwest of Garden City) in 1997. In 2000,
trapping and monitoring efforts were expanded to include Area II (southeast of Garden
City). Prior to 1970, these 2 areas were part of a contiguous tract of native sandsage
prairie. The development of center pivot irrigation led to the conversion of much of the
sandsage prairie to intensive agriculture (Waddell and Hanzlick 1978), and left these
68
areas as 2 fragments with about 19 km of agricultural fields between patch centroids
(Hullett et al. 1988, Jamison 2000).
Capture, marking, and monitoring
Lesser prairie-chickens were captured on 20 leks using walk-in funnel traps
(Haukos et al. 1991, Schroeder and Braun 1991) primarily during spring, and a subset of
leks during the fall of 1998 and 1999 for a related study (Salter and Robel 1999, Chapter
2). Spring trapping began the last week of March and continued into the second week of
April. Fall trapping lasted approximately 3 weeks in late September and early October.
Traps were placed on all known leks (>2 displaying males) found in native prairie on the
study sites. Once captured, birds were aged as yearling (~10 months of age) or adult (≥
22 months) based on shape, wear, and coloration of the 9th and 10th primaries (Amman
1944, Copelin 1963). Age was not recorded during the first year of the study (1997).
Gender of mature birds was determined by the patterns on tail and crown feathers
(Amman1944, Copelin 1963). Body mass was measured (± 2.5 grams) using a Pesola
spring scale.
Birds were individually marked with numbered aluminum leg bands.
Radiomarked birds were fitted with a lithium battery-powered necklace-style transmitter
(≤ 12-g), which was ≤ 1.9 % of a bird’s body mass (sexes pooled, x = 783, range = 630-
890 g) and below the recommended 3 % maximum level (Withey et al. 2001).
Transmitters had either an 8- or 12-hour mortality switch, and a 6- (1997-1999) or 12-
month (2000-2003) battery life. Nearly all captured females were marked with radio-
transmitters, but only a subset of males was marked in a given year. For these samples,
69
workers attempted to distribute the radios evenly across the study areas, by placing
roughly equal numbers on males on each lek trapped.
Radiomarked birds were monitored daily with a vehicle-mounted null-peak twin
Yagi antenna system. Once a mortality switch had been activated on a transmitter, the
unit was relocated and retrieved with a hand-held Yagi in < 12-hrs. Aerial surveys from
fixed wing aircraft were conducted at least once a month to locate undetected signals.
Once located by air, previously undetected birds that had emigrated from the primary
study areas were subsequently relocated approximately every 10 days from truck-
mounted systems.
Probable causes of mortality
When a carcass or kill site was located, the probable cause of mortality was
determined as mammal, avian, accident, hunter, snake, or unknown based on the
evidence at the recovery site. Scavenging of carcasses was possible but was potentially
minimized because of quick recovery of transmitters on a mortality pulse. It is difficult
to make unambiguous statements about causes of mortality because of potential
scavenging or confounding evidence at a recovery site (Bumann and Stauffer 2002).
Thus, I refer to probable causes of mortality in this paper.
A mortality event was classified as mammalian predation if one of the following
was evident at the kill site: bite marks on the transmitter, chewed feathers and/or
aluminum band in addition to tracks or mammalian scat at the recovery site. Coyote
(Canis latrans), American badger (Taxidea taxus), and bobcat (Lynx rufus) were all
possible mammalian predators. Carcasses that had been decapitated and/or cleaned of the
breast muscle with no apparent chewing, the presence of white wash, or plucking were
70
classified as avian predation. Accidental deaths were carcasses recovered near a
powerline, fenceline, or dismembered by farm machinery that had no apparent marks of
predation. Hunter recoveries of banded birds were reported as such. Often these were
not radiomarked birds, but are included here to provide an estimate of hunter harvest.
Snake predation was rare, but evidenced by feathers matted with saliva or residue from
the head to just above the furcula. Unknown causes included mortalities that went
undetected for more than 2 days, carcasses with confounding signs at the recovery site
(e.g., mammalian chew marks on feathers, etc., and under a powerline), or in the event a
carcass was recovered unscathed. Likelihood-ratio tests (G2) were used to examine the
differences in mortality causes between males and females.
Survival modeling
Known-fate survival modeling requires that 6 assumptions be met: the animals
marked are a random sample from the population, the experimental units (e.g., marked
animals) are independent, observation periods are independent, working radios are
always located, right-censoring is random, and that radios do not impact survival
(Winterstein et al. 2001). Right-censoring occurs when an individual “leaves” a study
either temporarily or permanently and the fate of the individual is unknown.
I must assume that captured females were a random subset of the population as all
captured females were marked. This assumption holds if nearly all females attend leks in
the spring for breeding. Males were marked so that an equal number from each lek were
monitored. To reduce bias in sampling, all leks were trapped on our study areas over a
relatively short period. Birds were tracked daily to enhance the detection probability of
functioning radios and transmitters with a mortality pulse (detection distance was ~1.6
71
km). Although an individual may have been missed for a day or so, birds that had not
permanently emigrated were found within a weekly sampling period. I modeled survival
with time units ≥ 1 week as these were more likely to be independent given that I
estimated survival over 9- and 12-month periods. Month was the base time unit for the
12-month estimates as aerial surveying was conducted monthly from September 2002
through March 2003. Birds were right-censored from the encounter histories if they were
known to emigrate or if a radio signal was lost. A 2-year comparison of survival rates of
radiomarked and banded birds from our study area found no measurable effect of
radiomarking (Chapter 2). I used a staggered entry known-fate encounter history and
included individuals after a 2-week acclimation (Winterstein et al. 2001) to the radio-
collar. Most birds would have entered the encounter history once our trapping efforts had
ceased. In fact, most mortality that occurred during the trapping season was excluded
from the analysis.
I estimated survival rates using known-fate models in MARK 3.0 (White and
Burnham 1999). Models were developed using the design matrix feature and a logit link
function. Global models were the most highly parameterized model, but not necessarily
saturated models. Model goodness-of-fit was evaluated by examining residual plots.
Model selection was based on the minimization of Akaike’s Information Criterion
corrected for small samples sizes (AICc), and AICc weights (wi’s) to select the model best
supported by the data. I used multi-model inference and model averaging in cases where
the difference in AICc values (∆AICc) between the best and subsequent models was < 2
(Burnham and Anderson 1998). Model averaging was used to refine parameter estimates
from candidate sets with several competing models. In this procedure, parameter
72
estimates (i.e., survival and standard errors) are calculated as a weighted mean using
AICc weights.
Data sets and candidate models
Transmitters in this study had different expected battery life; those used from
spring 1997 through fall 1999 had maximum life of 6 to 9-months (hereafter summer),
and those used from 2000 to 2003 had approximately a 12-month maximum life
(hereafter seasonal) (Table 1). I used the summer data to increase sample size of cohorts
across years (Table 1). However, these data may be biased if used to extrapolate to 12-
month survival estimates as they exclude winter survival. The bias could be positive if
winter survival is appreciably greater than the previous 8 months. Therefore, I analyzed
the seasonal data as a verification of the results from the summer analyses. In both sets
of analyses, data were pooled across various cohorts (Table 1) in order to have sufficient
sample sizes (n ≥ 25) to detect differences in timing and overall survival rates across
groups (Winterstein et al. 2001). Admittedly, such pooling may weaken the power of the
analyses, as there could have been confounding factors.
RESULTS
In total, 781 lesser prairie-chickens were captured during the 6-year study. Of
these, 136 males (46 yearling, 66 adult, 24 age undetermined) and 227 females (87
yearling, 117 adult, and 23 age undetermined) were fitted with radio-transmitters (Table
1). Fifteen birds (4%) (8 males and 7 females) died within the 2-week adjustment period
after marking. Birds were right-censored (30.8 %; 112 of 363 radiomarked individuals)
because of collar slippage and radio failure (79.5 %; 89 of 112), and emigration (20.5 %;
23 of 112) to other sagebrush fragments.
73
Summer survival
Monthly variation in survival was an important structure to all candidate sets, as
either an additive or a main effect (Table 2). Much of the time dependent variation was
centered on the nesting season (Fig. 1). Yearly variation in female summer survival was
well supported by the data (wi = 0.60), as summer survival (Ŝ) varied from 0.537 to 0.866
between 1997 and 2002 (Table 3). Age-specific survival of females with an additive
monthly effect (Fig. 1) was reasonably well supported by the data (wi = 0.54); however,
monthly variation alone fit nearly as well (∆AICc = 0.74; wi = 0.38) (Table 2). Yearling
females survived at a slightly higher rate (Ŝ = 0.757, SE = 0.041) than adult females (Ŝ =
0.692, SE = 0.041) during the summer (Table 3). Overall the monthly timing of mortality
(Fig. 2) and survivorship for males (Ŝ = 0.692, SE = 0.050) and females (Ŝ = 0.666, SE =
0.053) were similar, as time dependent and constant survival models fit better than those
with gender effects. There were several competing models in this candidate set, as the
top 5 models all had ∆AICc < 2 and wi‘s ranging from 0.13 to 0.29 (Table 2). However, a
plot of biweekly survival for each gender (Fig. 2) indicated that female survivorship was
inversely related to the cumulative number of females incubating nests, and increased as
nesting activities ceased. Alternatively, male survivorship appeared to decrease
periodically during the breeding season (Fig. 2).
Seasonal survival
Monthly variation was also important in model selection of the seasonal data, as
month was in the top 2 models of each candidate set (Table 4). Gender- and time-
specific survival was the most parsimonious model; however, the timing of mortality may
have been more important (Fig. 3) as the probability for males (Ŝ = 0.39 SE = 0.10) to
74
survive 12-months was similar to that of females (Ŝ = 0.43 SE = 0.05) (Table 5). Age-
specific survival with an additive time effect for females was well supported; however,
monthly variation alone was a highly comparable model (w1/w2 = 1.2; Table 4). The
probability of yearling females surviving the interval was substantially higher (Ŝ = 0.52
SE = 0.08), than that of adults (Ŝ = 0.37 SE = 0.06), but the difference in these
probabilities estimated from model averaging was less than non-averaged estimates
(Ŝyearling = 0.48 SE = 0.06; Ŝadult = 0.40 SE = 0.05)(Table 5). Monthly variation in
survival (Fig. 3) indicated that the greatest mortality occurred during the reproductive
season (Ŝ for Apr – Jul = 0.69, SE = 0.04), but winter survival (Nov – Feb) was not much
higher (Ŝ = 0.77, SE = 0.06).
Probable causes of mortality
One-hundred and thirty four mortalities were recorded from spring 1997 to spring
2003 (Table 6). The majority of mortality was attributed to mammalian predation (54%).
It is likely that coyotes were the primary mammalian predator, as most sign at recovery
locations suggested canid predation. Five whole carcasses were cached under shrubs, but
could only be classified as mammalian predation. Most of the mammalian predation
(75%) was associated with female mortality (G2 = 4.96, df =1, P = 0.004) and coincided
with losses during nesting (>30%).
Raptor predation occurred primarily during spring and winter, which coincided
with presence of wintering or migratory populations of red-tailed hawk (Buteo
jamaicensis), ferruginous hawk (Buteo regalis), rough-legged hawk (Buteo lagopus),
northern harrier (Circus cyaneus), and a few prairie falcons (Falco mexicanus). Great-
horned owls (Bubo virginianus) were present year round and were probable predators;
75
however, with no raptor predation in summer it is thought that their role as a predator was
minimal. Males seemed more susceptible to raptor predation (20%) than females (11%),
but this rate was not measurably different (G2 = 1.88, df =1, P = 0.170). Seven of the 8
accidents were associated with powerline collisions, and one female was killed by farm
machinery while incubating a nest in an alfalfa field. Snakes were responsible for the
mortalities of 3 nesting females. These females were > 20 days into incubation when
they were killed, and were likely tenacious in defending the nest. The gopher snake
(Pituophis catenifer) was the probable predator as this species was observed depredating
the eggs at 4 nests, and is the only snake on the study area large enough to constrict and
attempt to swallow the head and neck of a female. Losses to recreational hunting were
small (5%) relative to the radiotelemetry population, but even less (1.2% of 655 birds)
when compared to banded birds (781[total banded]−126[known mortalities] = 655) that
were available for harvesting. Males apparently were more susceptible (15%; 7 of 44) to
hunting losses than females (1%; 1 of 90) (G2 = 8.74, df =1, P = 0.003)
DISCUSSION
Yearly variation in summer survival was evident and seemed to be highly related
to mortality rates of incubating females on nests (Fig. 1); >30% of all female mortality
was associated with the nesting period. Summer survival rates and apparent nest success
(number of nests hatching >1 egg / total number of nests) from 5 years revealed the
strength of this relationship (Fig. 4). Changes in predator communities and/or residual
cover may have influenced nest success and/or female losses (Bergerud 1988), but such
factors were not measured in this study. However, management efforts that would
increase nesting success may markedly increase annual female survival.
76
Gender-specific survival
Gender-specific survival was not well supported by the data, as the overall
survival rates of males and females were similar in the summer and seasonal analyses.
However, the timing of survival over 12 months suggested that seasonal patterns in male
survivorship do not match the temporal pattern seen in females. The conspicuous
displays (or its physiological costs) of males may increase their vulnerability in both
spring and fall thereby equalizing the female losses during nesting (Amman 1957). This
result contradicts Bergerud’s (1988) prediction that uniparental care will result in greater
mortality in females than males.
If sexually selected characters increase mating success (e.g., conspicuous plumage
or behavior), then characters that increase survival (e.g., cryptic plumage or behavior)
should make a relatively small contribution to total fitness (Angelstam 1984). In grouse,
this should result in a lower survival rate in males (the gender with sexually selected
traits in this case) than females. One might expect this pattern to be evident in the most
sexually dimorphic species, capercaillie (Tetrao urogallus), greater sage-grouse
(Centrocercus urophasianus), blue grouse (Dendragopus obscurus), and black grouse
(Tetrao tetrix). Several studies have examined annual survival rates in male and female
grouse and the results have been mixed. Greater sage-grouse are the only species in this
group in which males have lower survival rates than females (Zablan et al. 2003). Less
sexually dimorphic species, white-tailed ptarmigan (Lagopus leucurus) and spruce grouse
(Falcipennis falcipennis) also exhibit gender-specific patterns of survival (Choate 1963,
Braun 1969, Ellison 1974, Keppie 1979), but the white-tailed ptarmigan are socially
77
monogamous. However, capercaillie exhibit the inverse with males having markedly
higher survival in 4 regions of Europe (Storch 2001), and male black grouse survival is ≥
to that of females (Angelstam 1984, Caizergues and Ellison 1997). Survival rates of male
blue grouse, greater prairie-chicken (Tympanuchus cupido), sharp-tailed grouse
(Tympanuchus phasianellus), lesser prairie-chicken, hazel grouse (Bonasa bonasia),
ruffed grouse (Bonasa umbellus), and rock ptarmigan (Lagopus lagopus) are nearly equal
to those of females (Weeden 1965, Elliot and Bendell 1967, Gullion and Marshall 1968,
Hamerstrom and Hamerstrom 1973, this study).
The lack of a consistent pattern across studies makes it difficult to draw
conclusions about the role or presence of gender-specific survival rates, and to account
for markedly different investments in breeding strategies between males and females
(Bergerud 1988). The relatively equal survival rates of male and female lesser prairie-
chickens suggest that the cost of reproduction has balanced female losses during nesting
with male mortality during display and at other times of year. Alternatively, the similar
plumages between males and females may equalize the survival rates when males are not
displaying. Additionally, these comparable survival rates contradict the idea that male
biased sex-ratios are a result of differential survival (Amman 1957, Campbell 1972, Moss
1987, Linden 1981, Bergerud 1988), suggesting that such ratios may reflect a bias for
males to be susceptibile to hunter harvest. While speculative, this conclusion is
supported by the fact that 7 of 8 hunter harvested birds in this study were males. Taylor
and Guthery (1980) documented that male lesser prairie-chickens visited grain fields in
flocks, and females did so in singles. Because the common method of hunting prairie
78
chickens involves pass-shooting flocks coming into grain fields, there is some support for
this claim.
Age-specific survival
Age-specific patterns in female survival were evident in both analyses with
yearling birds surviving at a higher rate than adults. This was surprising given that
yearlings of other grouse species tend to have reduced survival and reproductive output
(Bergerud 1988, Saether 1990). However, the timing of mortality was consistent
between these groups as models with an additive time effect (Sage + month) were the most
parsimonious in both sets of analyses, with the lowest survival rates during the nesting
and brooding periods. One might predict that the cost of reproduction would result in
reduced survival rates of adults, if they had higher reproductive output. However,
reproductive parameters were similar between age-classes in this study (Fig. 5), except
that yearling nest success, 33% (SE = 5 %), was marginally higher than adults, 28% (SE
= 4%). Whereas rates of nest loss were comparable, mortality rates were not. The reason
for such differences suggests different responses to depredation events on adults and
yearlings. Adult females may have been more tenacious (i.e., a fight response) in nest
guarding (see Hannon and Smith 1987), whereas the yearlings may have had a flight
response to nest predators. If true, this could explain the lower survival rate for adult
females.
Probable causes of mortality
Predator classification from evidence at kill sites may be problematic, because
mammalian scavengers may be most abundant in a region (Bumann and Stauffer 2002).
Overestimation of mammalian-kills increases as a function of ambient temperature, and
79
could have occurred in this study. However, if a positive bias occurred in this study it is
likely that it was consistent for both males and females. Thus, the magnitude of the
difference in cause specific mortality rates should be a reasonable index to compare
proportional losses between males and females.
The probable causes of mortality lesser prairie-chickens differed somewhat
between males and females, as 60% of predation on females was classified as
mammalian, 1.5 times as much as males (40%). Conversely, male losses (20%) were
classified as raptor predation more so than females (11%). This potential difference is
likely from the types of predators cueing on breeding and nesting behaviors of males and
females, respectively. Northern harriers have been documented harassing and killing
prairie grouse (Berger et al. 1963, Toland 1985, Haukos and Broda 1989). The
susceptibility of males to mortality may not occur at the lek site (Berger et al. 1963), but
the physiological cost of display (Vehrencamp et al. 1989) may increase the probability
of predation both during and after breeding season (Angelstam 1984). Mammalian
predation of displaying birds is probably rare (Hamerstrom et al. 1965), and this pattern
was likely true in my study. Spring and fall raptor migrations coincided with display
activities of lesser prairie-chickens in this study, thus likely increasing their susceptibility
to raptor predation.
In 2001, 2 females died of unknown causes during the nesting season. Each
carcass was recovered intact and with a brood patch (i.e., an area of the ventral surface of
the abdomen that a female will pluck to line the nest with feathers). One hen had been
incubating for 12 days at the date of death. The other had not been located on a nest
although her radiolocations suggested she had been laying eggs for ≥ 8 days. These birds
80
were submitted for necropsy at the Diagnostic Laboratory, Kansas State University,
College of Veterinary of Medicine. Avian cholera (Pasturella mutlicoda), was isolated
from various tissues of 1 female, and was suggested to be the cause of death.
Accidental death due to powerline and/or fence collisions has a substantial impact
on populations of European grouse (Miquet 1990, Bevanger 1995, Moss et al. 2000).
Powerline collisions occurred as birds were leaving the prairie to forage in adjacent
agricultural fields. This loss appeared to be relatively small in my study population,
accounting for 5% of all mortality, similar to rates estimated for greater sage-grouse in
Idaho (Connelly et al. 2000). It is possible that some portion of the “unknown” category
was associated with powerline collisions, thus 5% may be a conservative estimate.
Powerline densities were relatively low on the study area as they usually traversed a
partial length of the prairie edge (< 40 % of 100 km of edge).
Hunting mortality was low in this study of radiomarked birds (≤ 5%) and even
less for all birds (<1.5 %). Research on European grouse suggests that hunting mortality
of <10% should not impact spring breeding populations (Ellison et al. 1988, Ellison
1991a, b). Ellison (1991a) suggested that even declining populations can sustain some
harvest (~5%) without negatively impacting production. Alternatively, Small et al.
(1991) found that hunting mortality of 20% was mostly additive to both adult and
juvenile ruffed grouse. Connelly et al. (2000) suggested that harvest of females was
additive to over-winter mortality, but that some level of exploitation (~10 %) was
tolerable given the high annual variability in hunter success. Recreational hunting losses
appeared to have minimal impact on the study population in reducing either annual
survival or numbers of breeding birds in the spring.
81
MANAGEMENT IMPLICATIONS
Management of lesser prairie-chickens should focus on habitat manipulations that
increase nest survival and decrease female losses during incubation. In fragmented
populations, predator control may achieve short-term goals of increased female survival
and nest success, but may be too costly to sustain in the long-term (Schroeder and
Baydack 2001). Long-term management should focus on creating nesting habitat that
decreases losses to mammalian predation. Similarly, experimental work is needed to
understand the mechanisms by which mammalian predators search and find nests.
Specifically identifying habitat structure that lowers the success of nest predators on nests
and on females would be most advantageous for long-term habitat management.
Limiting harvest probably would have little impact on the lesser prairie-chickens in
Kansas since most mortality occurs during the nesting season. However, more accurate
information is needed on harvest rates throughout lesser prairie-chicken range. Current
harvest levels are estimated indirectly through mail surveys to upland gamebird hunters,
who may not actually hunt prairie chickens. A free permit system specific to prairie
chicken hunters would facilitate a direct measure of hunter harvests.
ACKNOWLEDGMENTS
I thank J. O. Cattle Co., Sunflower Electric Corp., Brookover Cattle Co., and P. E.
Beach for property access. B. E. Jamison, C. G. Griffin, J. C. Pitman, G. C. Salter, T. G.
Shane, T. L. Walker, Jr., and T. J. Whyte assisted with trapping and radiomarking of
prairie chickens. B. E. Jamison was instrumental in developing the initial candidate set
of survival models. Financial and logistical support was provided by Kansas Department
of Wildlife and Parks (Federal Aid in Wildlife restoration projects, W-47-R and W-53-
82
R), Kansas Agricultural Experiment Station (Contribution No. 03-_______), Division of
Biology at Kansas State University, and Westar Energy Inc.
LITERATURE CITED
Amman, G. A. 1944. Determining the age and sex of pinnated and sharp-tailed grouse.
Journal of Wildlife Management 8: 170-171.
Amman, G. A. 1957. The prairie grouse of Michigan. Department of Conservation Game
Division. Lansing, Michigan, USA.
Angelstam, P. 1984. Sexual and seasonal differences in mortality of black grouse Tetrao
tetrix in boreal Sweden. Ornis Scandinavica 15: 123-134.
Bendell, J. F., and P. W. Elliot. 1967. Behavior and the regulation of numbers in blue grouse.
Canadian Wildlife Service. Report Series Number 4. Ottawa, Ontario, Canada.
Berger, D. D., F. Hamerstrom, and F. N. Hamerstrom. 1963. The effect of raptors on prairie
chickens on booming grounds. Journal of Wildlife Management 27: 778-791.
Bergerud, A. T. 1970. Population dynamics of the willow ptarmigan Lagopus lagopus alleni
L. in New Foundland 1955 to 1965. Oikos 21: 299-325.
Bergerud, A. T. 1988. Population ecology of North American grouse. Pages 578-685 in A.
T. Bergerud and M. W. Gratson, editors. Adaptive Strategies and Population Ecology
of Northern Grouse. Wildlife Management Institute, Washington DC, USA.
Bevanger, K. 1995. Estimates of population consequences of tetraonid mortality caused by
collisions with high tension power lines in Norway. Journal of Applied Ecology 32:
744-753.
Boisvert, J. H. 2002. Ecology of Columbian sharp-tailed grouse associated with Conservation
Reserve Program and reclaimed surface mine lands in northwestern Colorado. Thesis,
University of Idaho, Moscow, Idaho, USA.
83
Braun, C. E. 1969. Population dynamics, habitat, and movements of white-tailed ptarmigan
in Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA.
Bumann, G. B., and D. F. Stauffer. 2002. Scavenging of ruffed grouse in the Appalachians:
influence and implications. Wildlife Society Bulletin 30: 853-860.
Burnham, K. P. and D. R. Anderson. 1998. Model selection and inference: a practical
information-theoretic approach. Springer. New York, New York, USA.
Caizerigues, A., and L. N. Ellison. 1998. Impact of radio-tracking on black grouse Tetrao
tetrix reproductive success in the French Alps. Wildlife Biology 4: 205-212.
Caizerigues, A., and L. N. Ellison. 1997. Survival of black grouse Tetrao tetrix the French
Alps. Wildlife Biology 3: 177-186.
Campbell, H. 1972. A population study of lesser prairie-chickens in New Mexico. Journal of
Wildlife Management 36: 689-699.
Choate, T. S. 1963. Habitat and population dynamics of white-tailed ptarmigan in Montana.
Journal of Wildlife Management 27: 684-699.
Connelly, J. W., A. D. Apa, R. B. Smith, and K. P. Reese. 2000. Effects of predation and
hunting on adult sage grouse Centrocercus urophasianus in Idaho. Wildlife Biology
6: 227-232.
Copelin, F. F. 1963. The lesser prarie-chicken in Oklahoma. Oklahoma Wildlife
Conservation Department. Technical Bulletin No. 6. Oklahoma City, Oklahoma,
USA.
Ellison, L. N. 1974. Population characteristics of Alaskan spruce grouse. Journal of Wildlife
Management 38: 383-395.
Ellison, L. N. 1991a. Shooting and compensatory mortality in tetraonids. Ornis Scandinavica
22: 229-240.
Ellison, L. N. 1991b. Under what conditions can shooting of declining species of tetraonids
84
be justified in France? Gibier Faune Sauvage 8: 353-365.
Ellison, L. N., P. Leonard, and E. Menoni. 1988. Effect of shooting on a black grouse
population in France. Pages 117-128 in M. Spagnesi and S. Toso, editors. Acts of
First National Conference of Game Biologists.
Gullion, G. W., and W. H. Marshall. 1968. Survival of ruffed grouse in a boreal forest. The
Living Bird 7: 117-167.
Hamerstrom, F., D. D. Berger, and F. N. Hamerstrom. 1965. The effect of mammals on
prairie chickens on booming grounds. Journal of Wildlife Management 29: 536-542.
Hamerstrom, F. N., and F. Hamerstrom. 1973. The prairie chicken in Wisconsin. Wisconsin
Department of Natural Resources. Technical Bulletin No. 64. Madison, Wisconsin,
USA.
Haukos, D. A., and G. S. Broda. 1989. Northern harrier predation of lesser prairie-chicken.
Journal of Raptor Research 23: 182-183.
Haukos, D. A., L. M. Smith, and G. S. Broda. 1990. Spring trapping of lesser prairie-
chickens. Journal of Field Ornithology 61: 20-25.
Hulett, G. K., J. R. Tomelleri, and C. O. Hampton. 1988. Vegetation and flora of a sandsage
prairie site in Finney County, southwestern Kansas. Transactions of the Kansas
Academy of Science 91: 83-95.
Jamison, B. E. 2000. Lesser prairie-chicken chick survival, adult survival, and habitat
selection and movement of males in fragmented landscapes of southwestern Kansas.
Thesis, Kansas State University, Manhattan, Kansas, USA.
Keppie, D. M. 1979. Dispersal, overwinter mortality, and recruitment of spruce grouse.
Journal of Wildlife Management 43: 717-727.
Lindén, H. 1981. Estimation of juvenile mortality in the capercaillie, Tetrao urogallus, and
the black grouse, Tetrao tetrix, from indirect evidence. Finnish Game Research 39:
85
35-51.
Merchant, S. S. 1982. Habitat use, reproductive success, and survival of female lesser prairie-
chickens in two years of contrasting weather. Thesis, New Mexico State University,
Las Cruces, New Mexico, USA.
Miquet, A. 1990. Mortality in black grouse Tetrao tetrix due to elevated cables. Biological
Conservation 54: 349-355.
Moss, R. 1987. Demography of capercaillie Tetrao urogallus in northeast Scotland. II. Age
and sex distribution. Ornis Scandinavica 18: 135-140.
Moss, R., N. Picozzi, R. N. Summers, and D. Baines. 2000. Capercaillie Tetrao urogallus in
Scotland-demography of a declining population. Ibis 142: 259-267.
Saether, B. E. 1990. Age-specific variation in reproductive performance of birds. Current
Ornithology 7: 251-283.
Salter, G. C., and R. J. Robel. 2000. Capturing lesser prairie-chickens on leks during fall.
Transactions of the Kansas Academy of Science 103: 46-47.
Schroeder, M. A., and R. K. Baydack. 2001. Predation and the management of prairie
grouse. Wildlife Society Bulletin 29: 24-32.
Schroeder, M. A., and C. E. Braun. 1991. Walk-in funnel traps for capturing greater prairie-
chickens on leks. Journal of Field Ornithology 62: 378-385.
Small, R. J., J. C. Holzwart, and D. H. Rusch. 1991. Predation and hunting mortality of
ruffed grouse in central Wisconsin. Journal of Wildlife Management 55: 512-520.
Toland, B. 1985. Northern harrier predation of greater prairie-chickens in southwest
Missouri. Journal of Raptor Research 19: 146-148.
Vehrencamp, S. L., J. W. Bradbury, and R. M. Gibson. 1989. The energetic cost of display
in male sage grouse. Animal Behaviour 38: 885-896.
86
Waddell, B., and B. Hanzlick. 1978. The vanishing sandsage prairie. Kansas Fish and Game
Magazine1-4.
Weeden, R. B. 1965. Breeding density, reproductive success, and mortality of rock ptarmigan
at Eagle Creek, central Alaska from 1960 to 1964. Transactions of the North
American Wildlife and Natural Resources Conference 13: 336-348.
White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from
populations of marked animals. Bird Study 46: 1-31.
Withey, J. C., T. D. Bloxton, and J. M. Marzluff . 2001. Effects of tagging and location in
error in wildlife radiotelemetry studies. Pages 45-75 in J. Millspaugh and J.
Marzluff, editors. Radio Tracking and Animal Populations. Academic Press,
San Diego, California, USA.
Winterstein, S. R. 1992. Chi-square tests for intrabrood independence when using the
mayfield method. Journal of Wildlife Management 56: 398-402.
Zablan, M. A., C. E. Braun, and G. C. White. 2003. Estimation of greater sage-grouse
survival in North Park, Colorado. Journal of Wildlife Management 67: 144-154.
87
Tabl
e 1.
Num
bers
of y
earli
ng (Y
) and
adu
lt (A
) les
ser p
rairi
e-ch
icke
ns ra
diom
arke
d in
Fin
ney
Cou
nty,
Kan
sas 1
997-
2002
.
M
ales
Fem
ales
Yea
r
Are
aI
Mon
ths
n (Y
,A)
a A
rea
II
n (Y
,A)
Mon
ths
Are
a I
n (Y
,A)
Mon
ths
Are
aII
n (Y
,A)
Mon
ths
1997
-199
8
24 (n
o-ag
e) b
9
ND
c N
D23
(no-
age)
b,d
9N
DN
D
1998
-199
9 34
(17,
17)
b
12h
ND
N
D
23
(18,
5) b
,d,e
9N
DN
D
1999
-200
041
(18,
23)
b12
h
ND
N
D
34
(18,
16)
b,d
,e9
ND
ND
2000
-200
1 N
D
ND
12
(5, 7
) g
12
26
(7, 1
9) d
,e,f,
g 12
25
(11,
14)
e,f,g
12
2001
-200
22(
0, 2
) g
12
13 (6
, 7) g
12
33 (1
2, 2
1) d
,e,f,
g 12
23
(8, 1
5) e
,f,g
12
2002
-200
3N
DN
DN
DN
D
14 (3
, 11)
d,e
,f,g
12
26 (1
0, 1
6) e
,f,g
12
Tota
ls
101(
35, 4
2 )
ND
25
(11,
14)
N
D
15
3 (5
8, 7
2)
ND
76
(29,
45)
N
D
88
a Max
imum
bat
tery
life
and
trac
king
per
iod
in m
onth
s for
a g
iven
coh
ort.
b Coh
orts
use
d to
exa
min
e se
x-sp
ecifi
c ra
tes f
or 9
mon
th p
erio
d; a
ges a
nd y
ears
poo
led.
c N
D =
no
data
ava
ilabl
e.
d Coh
orts
use
d to
exa
min
e an
nual
var
iatio
n in
surv
ival
rate
s for
9 m
onth
per
iod;
age
s poo
led.
e C
ohor
ts u
sed
to e
xam
ine
age-
spec
ific
rate
s in
fem
ales
for 9
mon
th p
erio
d; a
reas
and
yea
rs p
oole
d.
f Coh
orts
use
d to
exa
min
e ag
e-sp
ecifi
c ra
tes i
n fe
mal
es fo
r 12
mon
th p
erio
d; a
reas
and
yea
rs p
oole
d.
g Coh
orts
use
d to
exa
min
e se
x-sp
ecifi
c ra
tes f
or 1
2 m
onth
per
iod;
age
s, ar
eas,
and
year
s poo
led.
h T
wel
ve m
onth
s of c
over
age
was
pos
sibl
e du
e to
fall
trapp
ing
effo
rts.
Table 2. Candidate models and model statistics for summer survival (Apr-Nov) of lesser prairie-chickens in Finney County, Kansas, 1997-2002. Model statisticsa
Model structure
AICc ∆AICc wi K Dev
Annual variation (females only)
Syear + month 370.02 0.00 0.60 14 60.75
Syear + biweek 372.11 2.09 0.21 23 44.38
Smonth 372.30 2.28 0.19 9 73.20
Sconstant 399.08 29.06 0.00 1 116.09
Syear 399.10 29.09 0.00 6 106.06
Age-specific (females only)
Sage + month 454.86 0.00 0.55 10 21.43
Smonth 455.60 0.74 0.38 9 24.19
Sage + biweek 459.85 4.99 0.05 19 8.17
Sbiweek 460.44 5.58 0.03 18 10.79
Sage * month 468.51 13.66 0.00 18 18.87
Gender-specific
Smonth 353.01 0.00 0.29 9 38.97
Sconstant 353.65 0.65 0.21 1 55.72
Sbiweek 354.36 1.35 0.15 18 22.00
Ssex + month 354.45 1.45 0.14 10 38.39
Ssex 354.67 1.67 0.13 2 54.74
Ssex + biweek 355.81 2.81 0.07 19 21.41
Ssex * month 358.68 5.68 0.02 18 26.32
a Model fit is described with deviance (Dev), the number of parameters (K), and
Akaike’s Information Criterion corrected for small sample size (AICc).
89
Table 3. Summer survival estimates for radiomarked lesser prairie-chickens
after marking (Apr – Nov) 1997-2002. Parameter estimates
Group
Ŝ SE 95% CI
Year (females only)
1997 0.556 0.110 0.340, 0.772
1998 0.845 0.083 0.683, 1.000
1999 0.777 0.081 0.619, 0.935
2000 0.537 0.102 0.337, 0.737
2001 0.548 0.092 0.368, 0.729
2002 0.866 0.088 0.694, 1.000
Age-specific (females only)
Yearling 0.757 0.041 0.677, 0.836
Adult 0.692 0.041 0.612, 0.773
Gender-specific
Male 0.692 0.050 0.594, 0.790
Female 0.666 0.053 0.562, 0.771
90
Table 4. Candidate models and model statistics for seasonal (Apr-Mar) survival of lesser prairie-chickens in Finney County, Kansas, 2000-2003. Model structureb
AICc ∆AICc wi K Dev
Gender-specific
Ssex * month 543.72 0.00 0.61 24 17.16
Smonth 545.24 1.52 0.28 12 43.53
Ssex + month 547.15 3.43 0.11 13 43.39
Sconstant 556.34 12.62 0.00 1 76.92
Ssex 558.19 14.47 0.00 2 76.76
Age-specific (females only)
Sage+ month 455.91 0.00 0.55 13 429.51
Smonth 456.32 0.41 0.45 12 431.97
Sage*month 470.49 14.59 0.00 24 421.16
Sage 471.51 15.61 0.00 2 467.50
Sconstant 472.39 16.47 0.00 1 470.37 a Model fit is described with deviance (Dev), the number of parameters (K),
and Akaike’s Information Criterion corrected for small sample size (AICc).
91
Table 5. Seasonal survival estimates for radiomarked lesser prairie-chickens 12-months after marking (Apr – Mar) 2000-2003. Parameter estimates
Group
Ŝ SE(Ŝ) 95 % CI
Gender-specifica
Male 0.393 0.100 0.175, 0.630
Female 0.432 0.050 0.335, 0.533
Age-specific (females only)a
Yearling 0.481 0.066 0.379, 0.673
Adult 0.400 0.054 0.272, 0.504 a Parameter estimates were derived using the model averaging
procedure in MARK.
92
Table 6. Numbers and percentages of potential mortality causes of lesser prairie-chickens
in Finney County, Kansas, 1997 – 2003.
1997-1999 2000-2003 1997-2003
Potential cause
Femalea Male Total Femalea Male Total Overalla
Mammal 15 10 25 (61%) 39 8 47 (51%) 72 (54%)
Unknown 2 3 5 (12%) 17 4 20 (22%) 25 (18%)
Raptor 1 4 5 (12%) 9 5 14 (15%) 19 (14%)
Accident 1 1 2 (5%) 3 3 6 (6%) 8 (6%)
Hunterb 0 4 4 (10%) 1 2 3 (3%) 7 (5%)
Snake 0 0 0 3 0 3 (3%) 3 (2%)
Total 19 22 41 71 22 93 134
a Mortality during nesting was a considerable portion (32%) of all female losses 10 of 22
(46%) and 19 of 68 (27%) during 1997-1999, and 2000-2003, respectively.
b Harvest rates are biased high as reported birds were banded only, and not radiomarked.
93
Fig. 1. Annual variation in monthly survival rates (9-months) of females (A) from the
model S year + month, standard errors not included for clarity. Yearling and adult female
survival rates (B) from the model S year + month. Male and female survival (C) from the
model S gender*month (1997-1999).
94
0.90
0.92
0.94
0.96
0.98
1.00
YearlingAdult
Surv
ival
(4 w
eeks
)
0.88
0.90
0.92
0.94
0.96
0.98
1.00
199719981999200020012002
Apr May Jun Jul Aug Sep Oct Nov Dec
0.90
0.92
0.94
0.96
0.98
1.00
MaleHen
(A)
(B)
(C)
95
Fig. 2. Relationship between biweekly survival rates of males (dashed line) and females
(solid line) and the cumulative frequency distribution (CFD) of incubating females
(gray). Note the inverse relationship of female survival relative to the CFD of nests.
96
2-week intervals
Apr Jun Aug Oct Dec
CFD
(nes
ts)
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Kno
wn
fate
sur
viva
l est
imat
es
0.92
0.94
0.96
0.98
1.00
CFD (nests)Female survivalMale survival
97
Fig. 3. Monthly survival rates of males and females (A) for 12-months (2000-2003) from
the model S gender*month. Note the variation in the timing of survival between the groups.
Monthly survival rates for yearlings and adults (B) for 12-months.
98
Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar0.50
0.60
0.70
0.80
0.90
1.00
MaleHen
Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
Mon
thly
Sur
viva
l
0.80
0.85
0.90
0.95
1.00
YearlingAdult
(A)
(B)
99
Fig. 4. Nest success (black) and female survival (white) ± SEs for 1998-2002. Nests
from 1997 were not included due to different nest marking techniques.
100
1998 1999 2000 2001 2002
Surv
ival
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Nes
t Suc
cess
0.18
0.20
0.22
0.24
0.26
0.28
0.30
0.32
0.34
SurvivalNest Success
101
Fig. 5. Reproductive parameters of female lesser prairie-chickens (black =yearling; white
= adult) from 1998 to 2002. Note the consistency among clutch size, incubation date, and
probability of (re)nesting (A, B, and D) with only slight separation in nesting success (C)
between the age-classes.
102
Firs
t nes
tR
enes
t
Number of Eggs
0246810121416
Firs
t nes
tR
enes
t
Day of Year
100
120
140
160
Firs
t nes
tR
enes
t
Probability of nest success
0.0
0.1
0.2
0.3
0.4
(A)
(B) (D
)(C
)
Firs
t nes
tR
enes
t0.
0
0.2
0.4
0.6
0.8
1.0
Probability of (re)nesting
103
CHAPTER 4
LESSER PRAIRIE-CHICKEN DEMOGRAPHY: A SENSITIVITY ANALYSIS
OF POPULATION DYNAMICS IN TWO PRAIRIE FRAGMENTS
Abstract. Recently it has been suggested that nest success and chick survival are the
main limiting factors for populations of lesser prairie-chicken (Tympanuchus
pallidicinctus) in the sand sagebrush (Artemisia filifolia) prairie of southwestern Kansas.
This hypothesis was examined using elasticity analysis on an age-specific projection
matrix. The model was parameterized with demographic data from a 1998 to 2003 field
study of radiomarked lesser prairie-chickens near Garden City, Kansas. Additionally, the
projection matrices of two spatially implicit populations were compared to examine the
contributions of the vital rates to the difference in the rate of population change (λ)
between these fragments with contrasting human disturbance and sand sagebrush
communities. Lambda was less than 1.0 for both populations (λI = 0.544, λII = 0.754).
This indicated a decline in population growth in the absence of immigration. However,
the marked contrast in the contributions to λ between populations yielded differences in
sensitivity to various life-stages, and prescribed management practices on the two areas.
The application of this analysis to management of the sand sagebrush habitat could
possibly increase the effectiveness of management efforts.
104
INTRODUCTION
The lesser prairie-chicken (Tympanuchus pallidicinctus) is one of five prairie
grouse species that requires native grasslands and shrublands for breeding, nesting, and
annual survival. Historically this species’ range was one of the most limited,
concentrated in the mixed- and short-grass sand sagebrush (Artemisia filifolia) and sand
shinnery oak (Quercus havardii) prairies of southeastern Colorado, eastern New Mexico,
the panhandle of Texas, western Oklahoma, and southwestern Kansas (Giesen 1998).
This historic range has been further reduced and fragmented due to dry land agriculture
and more recently mechanized irrigation (Waddell and Hanzlick 1978). It is estimated
that over 90% of the range has been lost since the turn of the 20th century (Giesen 1998)
and the range-wide population decline has paralleled these losses. In 1995, the lesser
prairie-chicken was petitioned to be listed as threatened under the Endangered Species
Act (Giesen 1998). The U.S. Fish and Wildlife Service (USFWS) determined that such a
listing was “warranted but precluded” and this status is reviewed annually (USFWS
2002). Because of recent population declines, sport-hunting seasons were closed in New
Mexico in 1996 and in Oklahoma in 1998. Currently, a 2-day hunting season and 2/4
bird bag/possession limit occur in Texas, and Kansas has a 2-month hunting season and
1/4 bird bag/possession limit. Given the status of this sensitive species it is critical to
understand the dynamics of local and regional populations and identify possible
management scenarios that would most effectively benefit the lesser prairie-chicken.
An important concern for wildlife managers working with sensitive species is
knowing what constrains the population or which management practices will have the
greatest impact on the population (Peterson et al. 1998). Given the short duration of
105
research budgets it is often necessary to provide an adaptive framework for which
populations can be managed with short-term management experiments. That is, develop
a management hypothesis, implement the program, and assess the results of a given
experiment. Risk assessment models and population viability analyses (PVA) have been
used to provide insight to the management of sensitive wildlife species (Crouse et al.
1987, Wisdom and Mills 1997, Johnson and Braun 1999, Blakesly et al. 2001). Similar
approaches have been used in providing conservation guidelines for other grouse species,
including the two subspecies of the greater prairie-chicken (T. cupido pinnatus; Wisdom
and Mills 1997: and T. c. attwaterii; Peterson et al. 1998), sharp-tailed grouse (T.
phasianellus; Temple 1992), greater sage-grouse (Centrocercus urophasianus; Johnson
and Braun 1999), black grouse (Tetrao tetrix; Caizergues and Ellison 1997), capercaillie
(Tetrao urogallus; Grimm and Storch 2000), and willow ptarmigan (Lagopus lagopus;
Steen and Erikstad 1996). Generally, these analyses have concluded that recruitment was
the limiting factor affecting growth of populations with low adult survivorship. This
follows Bergerud’s (1988) hypothesis that nesting and brood rearing are the limiting
factors for grouse. Studies of longer lived species found that hunting mortality may have
been additive in years of poor recruitment (Steen and Erikstad 1996, Caizergues and
Ellison 1997, Johnson and Braun 1999). Few models were based on an ecological study,
and most required the use of surrogate parameter estimates from a closely related species,
age- or sex-ratios from harvest data, or previous publications of the study species.
This paper describes the development of an age-based matrix model of lesser
prairie-chicken population dynamics that compares the relative importance of the
reproductive and survival rates in two spatially implicit populations. The analyses in this
106
paper were based on a comprehensive ecological study from which vital rates were
estimated for each life history stage. I followed the modeling framework of Caswell
(2001) to address the various conservation concerns for this species and conducted the
following: 1) population assessment, 2) diagnosis, 3) prescription and 4) prognosis. I
assessed the lesser prairie-chicken populations by estimating the arithmetic rate of
population change (λ) and its respective confidence intervals to determine whether the
populations were stable (λ ≈ 1.0) or reflective of the long-term negative trend in these
populations (Jamison 2000). Retrospective analysis (or a Life Table Response
Experiment) was used to diagnose the probable causes of the population status. A
combination of prospective (e.g., elasticity and variance scaled sensitivity) and
retrospective (life-stage simulation analyses) techniques was used to prescribe
management strategies that were aimed at improving population status. Finally, I
returned to estimates of λ for both populations for the prognosis.
Less than 5% of radiomarked birds moved between these areas in a given year,
but emigration to other patches was 20%; estimates of various population parameters
indicated that area specific rates were warranted. Thus, I defined these two areas as
populations. The two study areas in southwestern Kansas differ in terms of human
disturbance, within-patch fragmentation, and density of sand sagebrush.
The specific objectives for each population were to 1) quantify the rate of
population change and examine whether the populations were stable, 2) compare the
relative importance of each vital rate to population status, i.e., test the nesting and brood
rearing limiting factor hypothesis, 3) identify which rate or combinations of rates have
107
the greatest management priority, and 4) determine what is the long-term viability of
these populations.
METHODS
Study Areas
The study region was comprised of two ~5,000 ha fragments of native sand
sagebrush prairie near Garden City, Finney County, Kansas. Prior to the 1970s, these
areas were a contiguous tract of sagebrush grassland (Waddell and Hanzlick 1978). The
development of center-pivot irrigation systems led to the conversion of 150,00 ha of sand
sagebrush to agricultural land. These areas were separated by ~20 km of center-pivot
irrigated fields of corn, alfalfa, and wheat. Vegetation in the prairie fragments was
primarily sand sagebrush, yucca (Yucca spp.), bluestem (Andropogon spp.), and big
sandreed (Calamovilfa gigantica). Soils were in the choppy sands range site category,
and topography was generally flat to rolling hills and dunes (Hullett et al. 1988). The
native rangelands were grazed seasonally by livestock. Several of the pastures on both
areas were treated with aerial applications of tebuthiron to control sagebrush densities.
These areas were similar in terms of overall type of habitat and land uses, but
varied by the quantities of each (Table 1). However, Area I had more linear km of road
(154 vs. 129 km), a higher density of pump-jacks (i.e., a device used in the extraction of
natural gas or crude-oil) density (10.6 vs. 6.9 ha-1,000), and more buildings (6 vs. 1; e.g.,
compressor stations, human dwellings) than Area II. Using the point-centered quarter
method (Cottam and Curtis 1956), sagebrush density, height and diameter were estimated
using 35 points (4 measurements per point) within a pasture (a discrete unit) and a
weighted mean density (accounting for area [ha] of each pasture) was calculated across
108
pastures (Area I = 10 pastures, x = 3,611, SE = 689;, Area II = 15 pastures, x = 4,206,
SE = 697) for each area (Table 1) .
Field Procedures
Lesser prairie-chicken females were captured on leks during the spring (1998 to
2002) using walk-in funnel traps (Haukos et al. 1990). Each captured female was
classified into an age-class yearling (≤ 10 mos) or adult (≥22 mos) by shape and wear on
the outer primaries (Amman 1947, Copelin 1953), marked with an aluminum band on the
tarsus, and fitted with a necklace-style lithium powered radio-transmitter with a mass <
12-g. Transmitters were ≤ 1.9% of a bird’s body mass ( x = 783, range = 630-855 g) and
below the recommended 3% level (Withey et al. 2001). Transmitters had either an 8- or
12-hour mortality switch, and 6-month (1998-1999) or 12-month (2000-2003) battery
life. All radiomarked lesser prairie-chickens were tracked daily and remotely with a
vehicle-mounted twin-Yagi null-peak telemetry system throughout the year to determine
reproductive and survival status. Nesting data, brood size, and predation events of adults
and post-fledgling chicks were determined from hand held telemetry systems.
The onset of incubation (± 2 days) was determined from radiotelemetry as
females were localized for ~10-day period and were found at the exact azimuths from
fixed locations over a 3-day period. Nests were visited once at the onset of incubation to
determine clutch size and once again later when the nest had either hatched or failed. If a
female was absent from her nest for more than one day then the nest was visited on the
second day to determine its fate.
Flush counts were used to estimate survivorship of broods to fledging, 34-days
post-hatch. Thirty-four days was chosen as the fledging date because chicks
109
demonstrated the ability to survive separately from their natal brood for up to 2-weeks.
Additionally chicks were first radiomarked at this age and known-fate survival was used
to estimate survival beyond that date. Initial brood size was estimated from the number
of hatched eggshells present in the nest. Broods were flushed at 14, 24 and 34-days post-
hatch to determine changes in brood size over this period. These systematic flush counts
were conducted at dawn when often the female was brooding the chicks.
Radiotelemetry was used to determine survival of chicks from fledging to first
breeding and of breeding females. Thirty-four day old chicks (weighing ~ 200-g) were
marked with a 2-g necklace style radio-transmitter with a 90-day battery life and were
tracked daily. At 60-days post-hatch radiomarked chicks (weighing ~ 400-g) were
recaptured and fitted with an adult size necklace style transmitters.
Demographic rates of lesser prairie-chickens
Clutch size (CLUTCH) was the total number of eggs laid in the nest (Table 2).
Eggs were laid on the ground in a shallow scrape ~ 15 cm in diameter. Since all nests
were found early in incubation the effects of partial clutch loss during laying would not
have been detected. Partial clutch loss occurred in 30% of successful nests with a median
of 2 eggs lost during incubation. Clutch sizes for both first and renests were used as
constants for ages and areas, because biologically predicted to vary in this species
(Giesen 1998), and practically there was little variation (Table 2).
Nest success (NEST) was estimated as apparent nest success, the proportion of
nests hatching at least one egg. Mayfield estimates were not used because nests were
found early in incubation and within 24-hrs of failure or hatch (Table 2). Rates of nesting
success were estimated separately for each age and area.
110
Renesting (RENEST) was the probability of a female laying a replacement clutch,
if her first nest failed. RENEST was calculated as the number of females laying renests
divided by total number of unsuccessful first nests. Renests were detected and fates
determined as described above. There were 5 cases in which the incubation date of a
female’s “first” nest was noticeably later (julian day > 143), than the mean (julian day =
127, SE = 8 days) for known first nests. In this case, it was assumed that these females
lost their first clutch during laying and that the only nest documented for such an
individual was a renest (Table 2). Small samples sizes precluded inputting these
estimates as age specific rates.
Daily survival rates of pre-fledging chicks (CHICK) was estimated using a
modified Mayfield estimator (Flint et al. 1995) applied to the change in brood size
between flush counts. The daily rate was raised to the power of 34 to estimate survival
over the 34-day period. This rate may be a conservative estimate if chicks were mixing
with other broods (especially unmarked) at any time during the 34-day post-hatch period.
Although, limited mixing (5%) was documented in radiomarked birds > 34 days of age
in this study, mixing prior to this fledging period has not yet been documented in
galliformes. Thus, the assumption was made that losses in brood size were deaths (Table
2). Small sample sizes precluded meaningful estimation between age-classes, but area
specific rates were calculated.
I estimated survival rates of 34-day old chicks to first breeding (PBS, P0) and
annual survival rate for females (HEN,P1, P2) using known-fate models in MARK 3.0
(White and Burnham 1999), and all models were developed using design matrices and
logit link functions. Model fit was assessed by examining the residual plots. Model
111
selection was based on the minimization of Akaike’s Information Criterion corrected for
small samples sizes (AICc), and AICc weights (wi’s) to select the model best supported by
the data. Post-brood survival was estimated as one rate for all cohorts because of small
sample sizes, but HEN survival was estimated for each cohort.
Population model
A deterministic female life-cycle model was constructed to summarize the age-
structured variation in vital rates for both populations (Fig. 1). In this pre-breeding birth-
pulse model, census occurred in April when birds were captured. The expected number
of female chicks produced per female (Fi) were
Fi = [(CLUTCH1 × NEST1) + (RENEST × CLUTCH2 × NEST2)] ×
(0.5 × 0.93 × CHICK)
where subscript, i, denotes the age-class, and the subscripts 1 and 2 indicate parameter
estimates associated with first nesting and renesting attempts, respectively. Two
adjustment factors were used to account for the assumed 1:1 sex ratio at hatch (0.5) and
egg hatchability (i.e., 1–[the proportion of infertile and unhatched embryos / total clutch
size]) which was fairly constant at 0.93 (SE = 0.01). Total reproductive output for a
given age-class (F1, F2) was combined with survival rates of juveniles (P0) (Fig. 1).
The arithmetic rate of population change (λ), stable age (w), reproductive value
(v), and sensitivity analyses were derived using algorithms (Caswell 2001) in
MATLAB 6.5 software (Mathworks Inc.). Confidence limits (95%) around the
observed λ, were calculated using the parametric bootstrap method (Manly 1997, Ebert
1999, Fieberg and Ellner 2001). Bootstrapped λ values were calculated as follows: 1)
resample lower-level vital rate from a probability distribution (i.e., beta-distribution for
112
probabilities, and normal distribution for continuous variables), 2) parameterize the
projection matrix of resampled rates, 3) calculate λ from the re-parameterized matrix, and
4) repeat 1,000 times. If λ =1 in the upper or lower 2.5% then the null hypothesis of
population stability was not rejected.
Retrospective analysis
A life table response experiment (LTRE) was used to quantify the contribution of
actual variation of these rates on the variability in λ between the two populations
(Caswell 1996). I use superscript ‘I’ and ‘II’ to indicate matrices or parameter estimates
from Areas I and II, respectively. In a fixed effect, single class design the mean matrix,
A(.), can be parameterized from two populations [A(.) = (A(I) + A(II)) / 2] and one can
evaluate the partial derivatives of the mean matrix. In order to quantify the lower-level
contributions of two matrices this ‘effect’ of treatment II on λ can be decomposed by,
ii
ii xxx
∂∂
−≈− ∑ λλλ )( IIIIII .
This type of LTRE assumes that A my be reparameterized in terms of the the lower-level
rates (xi) (Caswell 1996). In practice the difference in an elasticity value of a given rate
(aij) is multiplied by the sensitivity (sij ) value of the mean matrix. This results in
elasticity values (eij) that describe the contribution of or variation explained by a lower-
level vital rate to the observed rate of population change.
Prospective analyses
Prescriptions for targeting management actions were ranked using sensitivity and
elasticity analyses. Elasticity values were derived analytically to compare the
proportional effect of infinitesimal changes in a vital rate to λ. Elasticities (eij) of the
113
matrix elements are scaled (log) sensitivity values (sij) so that they sum to 1 (Caswell
2001),
ij
ij
ijij a
aa
e∂∂
=∂∂
=λ
λλ
)(log)(log .
However, matrix elements, F1 and F2, were comprised of several lower-level rates
(xi). The contribution of these rates to λ were evaluated by taking the partial derivatives
of the matrix (Caswell 2001),
xa
ax
xx ij
ji ij ∂
∂
∂∂
=∂∂ ∑
,
λλ
λλ
.
This gives the proportional change in λ from a proportional change in a lower-level rate
(x). I also used variance scaled sensitivities (VSS) as a measure of vital rate elasticity,
because scaling (i.e., arcsin-square root transform in this study) associated with VSS
allows for an evaluation of a given change in a vital independent of the actual parameter
value (Link and Doherty 2002). Link and Doherty (2002) define prospective analyses as
the functional dependence of λ on xi, and the mean/variance relation for any random
variable distributed on the unit interval suggests that VSSs based on the arcsine square
root transform are more appropriate for demographic probabilities (θ) where,
VSS = θ
)θ1(θ)]θ(sin2[
log1 ∂
∂
−=
∂∂
−
λλ
λ .
I report both standard elasticities and VSSs for comparative rankings to other studies.
Perturbation analysis
As a compliment to the elasticity matrix and lower-level contributions,
perturbation analyses were used to identify future management alternatives (i.e.,
prescriptions) that would have the greatest impact on the populations (Mills et al.1999).
114
The four most important vital rates, as determined from a ranking of the VSSs were
selected for the perturbation analyses. I took two slightly different approaches to
evaluate conservation alternatives. First, I used a life stage simulation analysis (LSA)
(Mills et al. 1999) approach where the mean of a vital rate was increased by 10, 20, and
30% and its variance reduced by 10%. This method assumes that current estimates of
vital rates are unbiased. Second, I set a specific management target for each vital rate,
and examined what effect that rate had on λ. This modified LSA requires that the mean
of a rate of interest increase by a fixed amount and simultaneously reduce variability in
the rate (~10% in this study). Simulations for both perturbations were conducted similar
to the parametric bootstrap for confidence limits, except that the vital rate under
assessment was drawn from a uniform distribution with its respective mean and range
(Wisdom et al. 2000), and all other rates were drawn from the appropriate probability
distribution. Under the management target simulations both NEST and CHICK were set
to have a probability of 0.3, and 0.4; and PBS and HEN were set to have a mean of 0.5,
and 0.6 and a range of ± 0.1. I constrained perturbations to the stated probabilities as
these rates are similar to those reported for the species and its congeners (Schroeder and
Robb 1993, Giesen 1998). The effectiveness of a proposed management action was
assessed by the proportion of bootstrap replicates in a simulation where λ ≥ 1. It was
assumed that most management of lesser prairie-chicken populations would not be age-
specific, and would affect all age-classes in a similar manner. Thus, in these simulations
for NEST and HEN both yearling and adult females were drawn from the same
probability distribution.
115
RESULTS
Vital Rates
Clutch sizes were markedly different between first (12.1 eggs) and renests (7.7
eggs) and a similar pattern was observed at both areas (Table 3). Nest success varied
between areas (Table 3) but most of the variability was accounted for by the greater
probability of first nests (14–41%) being successful than renests (12–14%). The
probability of renesting after nest loss was not equal between areas (Table 3). Daily
survival rates of 34-day old chicks was similar between the areas (Table 3), although
samples sizes were small. Known-fate survival of females varied by area and age-class
with the yearling to two-year old transition having 13-15% higher survival than older
birds (Table 3).
Projection Matrices
The parameterized matrices reflected the age-structure and differences in rates
between the two areas,
A(I) = 302.0429.0144.0289.0
A(II) = 438.0588.0284.0224.0
.
The arithmetic rate of population change was markedly different between the two
populations λ(I) = 0.544 (95% CL = 0.28, 0.85) and λ(II) = 0.754 (95% CL = 0.47, 1.00).
The confidence limits indicated that Area I was declining and Area II was not
significantly different from λ = 1.0. The dominant left and right eigenvectors of the
matrix determined the reproductive value (v) and stable-age distribution (w) of the
population, respectively. Adults comprised the majority of the stable-age distribution
116
vector (w) on Area I (0.66) and Area II (0.65). Adults on the Area II had a higher
reproductive value (v = 0.90) than Area I (v = 0.59).
Retrospective analysis
Using Area I as the reference population, the effect of area on λ was λ(II) − λ(I) =
0.21. There were positive contributions from advantages in HENy,a on Area II,
especially for the 2+ age-class (Fig. 2). There was also a positive contribution from the
advantage of NEST1a on Area II, and it was the largest contribution overall. There was a
negative contribution from an advantage for CHICK on Area I (-0.09) and was the 2nd
largest overall contribution. Thus, most of the variability in λ(II) − λ(I) can be explained
by higher nest success on Areas II, higher chick survival on Area I, and to a lesser extent
higher HENy,a on Area II.
Prospective analyses
Elasticity values of the projection matrices (upper-level elasticities),
E(I) = 285.0228.0228.0259.0
and
E(II) =364.0263.0263.0111.0
,
indicated that future changes in HENa would have the largest effect on λ. Apparently, the
overall future reproductive contribution of F1 on Area II was relatively small. Analytic
elasticity values for the lower-level rates suggested that λ was most sensitive to future
changes in the rates PBS, CHICK (both ranked first), and HENa (ranked second) in these
populations (Table 4). Ranked as third largest effect on rates of population change were
117
HENy (λ(II)) and NEST1y (λ(l)). Renesting contributed little to future changes in λ. The
top four ranks for VSSs were consistent with analytic elasticities except for Area II.
However, on Area I, NEST1y and HENy were not ranked in the top 4, but NEST1a was
included. Variance scaled sensitivities reprioritized the importance of e(PBS) =
e(CHICK) and each having the largest effect on λ, to CHICK having the single largest
contribution to changes in λ. Interestingly, VSS ranking reduced PBS to a rank of third
for both populations, and NEST1a ranking increased to second for Areas I. Conversely,
the ranking of HENa and NEST1a dropped to fourth for Area I and II, respectively.
Perturbation analysis
The perturbation analysis simulated management practices that would presumably
yield large changes in λ with respect to the targeted vital rate (Mills et al. 1999). The
lower-level VSSs indicated NEST1, CHICK, PBS, and HEN were the most important
rates with respect to changes in λ, and were used in the perturbation analyses.
Hypothetical management scenarios for NEST1 and CHICK were assumed to respond the
greatest from habitat management. Practically, variables such as residual cover
(Buhnerkempe et al.1984) and forb abundance are important to probability of success of
these two life stages (Riley and Davis 1993). Manipulations to PBS and HEN were
surrogates for limiting (+10%) or eliminating (+20 and 30%) the hunting season as this is
typical management for increasing annual survival rates of exploited populations
(although its effectiveness is largely untested). Alternatively, stochastic events and
landscape factors that affect natal dispersal could also impact PBS.
Absolute increases in vital rates had little impact on λ (Table 5). Neither
population had large increases in λ for changes in single rates, and < 19% of any
118
simulation had λ ≥ 1. Suggesting that relatively large changes in vital rates are required
to have any measurable effect on λ. Increases in PBS and NEST1 + CHICK had the
largest impact on both populations but neither treatment at + 30% was large enough to
maintain population stability more than 18% of the time. Future increases in PBS had the
greatest effect on λ(I) as was predicted by the elasticity values. Interestingly, the vital
rates CHICK and HEN, which had the highest rank in the VSS analysis with respect to
their effects on λ(II), responded poorly to reductions in harvest or other management
practices that might increase these survival rates. Management that simultaneously
manipulated both NEST1 + CHICK rates had far more effect on λ than did changes in
HEN.
The results of the simulated management scenarios (Table 6) complemented the
LTRE and to a lesser extent the lower-level elasticity analysis. The perturbations
indicated that λ(I) responded most to habitat management that increased NEST1 + CHICK
to 40% and relatively less to increases in PBS and HEN. However, if targeting a single
vital rate, HEN had the largest impact on λ(I). Management that simultaneously targeted
both NEST1 + CHICK rates at 40% had nearly twice effect on λ(I) (Table 6) than did
managing for HEN of 60% (66 and 100% increase of the current rates). Management
scenarios with respect to λ(II) indicated that CHICK = 40% had the largest single impact
on λ(II). HEN and PBS had a greater effect on λ(II) than did NEST1. It is not surprising
then that NEST1 + CHICK had the largest effect on λ(II).
119
DISCUSSION
Parameter estimates
This study yielded several important results about the demography of lesser
prairie-chickens and illustrated some of the more complementary sensitivity analyses
(Mills et al. 1999, Wisdom et al. 2000). I am confident of the estimates of female
survival, post-brood survival and nest success in this study, because radiomarking does
not appear to affect survival negatively (Chapter 2). High rates of censoring (> 20%) can
positively bias known-fate survival estimates when sample sizes are small (< 50 animals)
(Tsai et al. 1999). Right-censoring of adults and yearlings in this study was ~30%, and
may have biased my estimates. Alternatively, right-censoring of juvenile birds during
post-brood survival was low (6.3%) and those estimates are likely accurate, albeit from
32 birds over 3 years. Nest success in this study was similar to estimates in the literature
(Giesen 1998). Females were flushed only once during incubation to determine clutch
size. Westemeier et al. (1998) reported that flushing greater prairie chicken females from
nests resulted in 95% of those females returning to nests, and nest success was not
negatively impacted. It is possible that estimates of CHICK were biased low due to
systematic flushing; however, > 33% of all broods in this study suffered complete brood
loss prior to the first flush at 14-days post-hatch. Most flush-counts were conducted in
the morning while chicks were being “brooded” by the female, thus detection probability
was likely high. As chicks reached 34 days of age, their mobility and independence was
such that they could mix readily with other broods and it is possible that estimates of
CHICK were biased low later in the sampling period. Small sample sizes were the
greatest limitation in estimating this rate.
120
Age-structure was most prominent in survival rates when compared to the other
rates in the populations, as the transition from yearling to adult age-class was the highest
rate on both areas. This pattern of high yearling survival also occurred in banded males
during this study (Chapter 1), and reflected a life-history strategy of a “short-lived large
clutch size species” (Saether and Bakke 2000). Yearling males exhibited different
breeding behaviors in terms of site fidelity, which may have increased survival to a
second year. Similarly, yearling females may not be as tenacious in nest-defense leading
to less mortality during incubation (Chapter 3).
Projection matrix
Population assessment.—The stable age distribution (w) projected by the matrix
was similar to the actual age-structure of the radiomarked sample for both areas (yearling
= 0.38, adult = 0.62). Meeting the assumption of w in the model is important especially
when reproductive values (v) are unequal in the population (Hoekman et al. 2002), as was
the case in both of my populations. The λ’s projected 25-46% declines, but only Area I
was significantly different from λ =1. Age-ratios (0.27-0.786 yearling/adult) during
spring trapping suggests either recruitment from outside the study area or negative bias in
estimated rates of reproduction. The confidence limits around λ in both populations
suggested that in fact this rate was similar between areas. The large range of confidence
limits resulted because my parameter estimates also included sampling variance. Longer-
term studies can separate process and sampling variance and yield estimates that are more
precise. Although, the confidence bands around λ suggested no differences in these
populations, vital rates contributing to these observed values were quite different.
121
Past variability and LTRE
Population diagnosis.— LTRE has not been used previously in demographic
studies of grouse or in other conservation literature (Caswell 2001); thus comparisons
within or outside this taxon is not possible. Although a powerful tool, I was limited in
my analysis by having only 2 sites for comparisons. A larger number of study areas or an
experimental design in which one or both areas were manipulated (and pre- post-
treatment matrices were parameterized and compared) would have provided greater
resolution in the analysis.
The LTRE analysis has two important points. First, the area effect on λ was
relatively large (0.21) because fragmentation and habitat loss may have had an effect on
vital rates. These contributions were several orders of magnitude larger in the case of
NEST1 but less so for CHICK with respect to the area effect. Thus, statistically
insignificant differences in λ do not necessarily equate to equal dynamics within
populations (Caswell 1996, Johnson and Braun 1999). Second, decomposition of the
contributions to λ with respect to the lower-level rates specified the aspects of Fi that
were important to each population, CHICK, PBS, and NEST1. This provided some
insight as to why λ may not have been equal between populations. The habitat and
landscape features differed between these areas. Vegetation only differed slightly in
terms of overall sagebrush density; however Area II had 2 pastures with the highest
densities (>9,000 plants ha-1) and highest probability of nest success. Modeling of chick
survival indicated that moderate stands of sagebrush (4,000 – 6,000 plants ha-1) yielded
the highest daily survival rate (DSR) of chicks through 14-d post-hatch (Pitman 2003).
Forty percent and 10% of the pastures on Area I and II, respectively, had density
122
estimates within this range, suggesting that the elevated chick survival on Area I may
have been explained in part by the prevalence of optimal sagebrush density.
There was substantially more human disturbance (i.e., within patch
fragmentation) on Area I and may have resulted in reduced fecundity and survival
compared to Area II. Habitat fragmentation has an array of detrimental demographic
effects (Knick and Rotenberry 1995), some of which were potentially borne out in this
study. However, without the appropriate replicates, I can only discuss the implications
for these two sites. The higher density of pump-jacks, buildings, and length of road on
Area I yielded a more highly fragmented habitat (Table 1) that could have had negative
impacts on survival or reproductive output. The confounding factors of habitat structure
and landscape features will require future work to differentiate any causal relationships.
Past variation in vital rates may be important to understanding future changes to
the matrix (Wisdom et al. 1999). The LTRE in this study elucidated the vital rates in
each population that were contributing to the observed differences in λ. This suggests
that past differences in habitat or landscape may have contributed to these differences in
nest success and chick survival.
Prospective analyses (prescriptions)
Analytic perturbations.—Variance scaled sensitivities indicated that the
functional relationship of λ on CHICK was ≥ 2.1 times larger than any other rates for
Area II and nearly so (≥1.7) for Area I. I concur with the assertion of Peterson et al.
(1998) that the sensitivity to CHICK should have a greater population-level impact
because total brood loss for a female results in no contribution of young that year. Total
clutch loss of first nests may have a similar effect, but there is a probability of renesting.
123
The VSS of HENa and NEST1a indicated that future changes in these rates had the second
largest impact on λ(II) and λ(I), respectively. This may result in part from a higher
sensitivity of λ to changes in nest success of older birds, because >30% of all female
mortality occurs during the nesting season (Chapter 3). Thus, steps taken to ensure
increases in nest success will likely also increase female survival. Additionally, yearlings
have the highest survival rates going into the second breeding season where nest success
was highest, thus yielding a relatively high reproductive value (vII = 0.9) and sensitivity.
PBS was valued similarly in both populations having the third largest effect on λ. The
importance of PBS to these populations may be two-fold: 1) it is an index to the
limitations of the actual recruitment rate, and 2) if the observed λ’s do indicate declining
populations, then PBS may also reflect the sensitivity of λ to immigration from other
habitat patches. First, Peterson and Silvy (1994) found that reproductive success could
predict the numbers of prairie chickens on leks the following spring, and Wisdom and
Mills (1997) found the survival of the first age-class was most important to simulated
population trends. My findings support these suggestions and perhaps provide a
generalized life-history strategy for prairie chickens, a life-history strategy of “boom or
bust” fecundity. Such a strategy may drive short-term dynamics in a short-lived large
clutch size organism. Second, PBS of radiomarked juveniles in my study is merely a
survival rate (S). However, at the population level PBS is comprised of two rates, S and
immigration or dispersal from other habitat patches. Because natal dispersal lesser
prairie-chicken often occurs just prior to the breeding season (Pitman 2003), new
individuals could be added to the population without experiencing the extrinsic factors of
either of my study areas. Given the declining λ’s of these populations, it is not surprising
124
that the vital rate that has a component somewhat independent of local constraints (and
directly adding to the population size) would be projected to have a relatively large
impact on population stability.
Comparisons to other grouse species.—Previous sensitivity analyses on 6 grouse
species have yielded mixed results (Table 7 and 8). Upper-level elasticity analyses
tended to rank survivorship of older birds higher than fecundity values (Table 7). This
was surprising for Attwater’s and greater prairie-chickens as their clutch size is markedly
larger (13 vs. 8 eggs) than that of the longer-lived greater sage-grouse and black grouse.
However, survival of juveniles to the first breeding (PBS in this study) was consistently
ranked highly among upper-level elasticities. This reiterates the importance of
recruitment both in terms of local survival and probability of immigration of the juvenile
age-class. I examined and ranked 4 lower-level elasticities that were comparable (e.g.,
nest success, chick survival, post-brood survival, and adult survival) across 4 studies
including my own (Table 8). This summary revealed that chick and PBS were generally
the first and second most important rates contributing to λ, respectively. This was
surprising given that capercaillie and willow ptarmigan have relatively small clutch sizes
compared to prairie grouse, because one would predict that λ would be most sensitive to
changes in adult survivorship in these longer-lived species.
Simulated perturbations.—Discrepancies between elasticity and perturbation
analyses was expected to some degree based on an intensive simulation study (Mills et al.
1999). Mills et al. (1999) provided evidence that infinitesimal and proportional changes
in a matrix do not reflect real changes in the environment of a species. They argue that
environmental stochasticity or management plans can potentially generate disproportional
125
and large changes in several rates simultaneously, yielding different inference about vital
rates of a population than elasticity analysis. Several studies (Mills et al.1999, Heppell et
al. 2000, Link and Doherty 2002, Nichols and Hines 2002) have concluded that elasticity
analysis was inadequate as a stand-alone result and incorporating other sensitivity indices
may yield a more complete understanding of the system. The lower-level elasticities
were weighted heavily towards adult survival and post-brood survival in this study. This
may have been due in part that the observed λ was <1 and declining populations tend be
more sensitive to adult survivorship. The perturbation analysis provided evidence in the
opposite direction indicating that nesting and brood rearing aspects of the life history
were critical to maintaining population stability for lesser prairie-chickens in southwest
Kansas. The importance of these early life stages was echoed in the LTRE as mentioned
above. Thus, increases in adult survivorship or post-brood survival by eliminating
hunting (<5% of all current mortality) would do little to stabilize these populations.
However, hunting of a declining population can only be justified if it is compensatory to
annual survival. This was contrary to the findings of Steen and Erikstad (1996) and
Johnson and Braun (1999); both studies found that decreases in adult survivorship from
hunting negatively affected population growth, especially in years with poor recruitment.
However, the markedly higher harvest rates in those studies likely had a larger effect on
survival than that observed in my study.
Conclusions
Population prognosis.— Ascertaining the status of these populations was difficult
given that both sampling and process variance were included in the parameter estimates.
However, efforts to increase nesting success and chick survival are paramount, and the
126
former likely will increase female survival. The tenuous stability of λ in these two
populations suggests that these fragmented populations are maintained in part by
immigration. Lek survey routes through this region have not detected the declines
reported here (KDWP unpublished data), and age-ratio data collected during this study
support the idea that immigration is a potential mechanism for population stability. It is
noteworthy that the results of this comprehensive field study reflected the conclusions of
other workers on prairie grouse (Peterson and Silvy 1994, Wisdom and Mills 1997,
Peterson et al. 1998) that extracted rates from the literature and across large geographical
regions. It is clear from this study and other modeling exercises (Wisdom and Mills
1997, Peterson et al. 1998) on prairie grouse that nesting and brood rearing are the critical
rates to maintaining population stability. This study also indicated the importance of
immigration via post-brood survival from other habitat patches. Management efforts
aimed at increasing the quality, and/or quantity of nesting and brood habitat will have the
greatest benefits to lesser prairie-chicken populations. Alternatively, this study provided
some initial insight as to the impacts of hunting on a declining species, and indicated that
eliminating hunting (if additive) would not yield a stable population.
ACKNOWLEDGMENTS
I thank J. O. Cattle Co., Sunflower Electric Corp., Brookover Cattle Co., and P. E.
Beach for property access. I thank J. C. Pitman for sharing his unpublished data on nest
success and chick survival, C. G. Griffin, B. E. Jamison, J. C. Pitman ,G. C. Salter, T. L.
Walker, Jr., and T. J. Whyte for assistance with field work. B. K. Sandercock provided
assistance with computer programming and advice on developing the matrix models.
Financial and logistical support was provided by Kansas Department of Wildlife and
127
Parks (Federal Aid in Wildlife restoration projects, W-47-R and W-53-R), Kansas
Agricultural Experiment Station (Contribution No. 03-_______), Division of Biology at
Kansas State University, and Westsar Energy Inc.
LITERATURE CITED Amman, G. A. 1944. Determining the age of pinnated and sharp-tailed grouse. Journal of
Wildlife Management 8: 170-171.
Bergerud, A. T. 1988. Population ecology of North American grouse. Pages 578-685 in
A. T. Bergerud and M. W. Gratson, editors. Adaptive Strategies and Population
Ecology of Northern Grouse. Wildlife Management Institute, Washington DC,
USA.
Blakesly, J. A., B. R. Noon, and D. H. Shaw. 2001. Demography of the California
spotted owl in northeastern California. Condor 103.
Buhnerkempe, J. E., W. R. Edwards, D. R. Vance, and R. L. Westemeier. 1984. Effects
of residual vegetation on prairie-chicken nest placement and success. Wildlife
Society Bulletin 12: 382-386.
Caizerigues, A., and L. N. Ellison. 1997. Survival of black grouse Tetrao tetrix the
French Alps. Wildlife Biology 3: 177-186.
Caswell, H. 1996. Analysis of life table response experiment II. alternative
parameterization for size and stage structured models. Ecological Modelling 88:
73-82.
Caswell, H. 2001. Matrix population models. Sinauer, Sunderland, Massachusetts.
Copelin, F. F. 1963. The lesser prarie-chicken in Oklahoma. Oklahoma Wildlife
Conservation Department. Technical Bulletin No. 6. Oklahoma City, Oklahoma,
USA.
Cottam, G., and J. T. Curtis. 1956. The use of distance measures in phytosociological
sampling. Ecology 37: 451-460.
128
Crouse, D. T., L. B. Crowder, and H. Caswell. 1987. A stage-based population model for
loggerhead sea turtles and implications for conservation. Ecology 58: 1412-1423.
Ebert, T. A. 1999. Plant and animal populations: methods in demography. Academic
Press. San Fransico, California.
Fieberg, J., and S. P. Ellner. 2001. Stochastic matrix models for conservation and
management: a comparative review of methods. Ecology Letters 4.
Flint, P. L., K. H. Pollock, D. Thomas, and J. S. Sedinger. 1995. Estimating prefledging
survival: allowing for brood mixing and dependence among brood mates. Journal
of Wildlife Management448-455.
Giesen, K. M. 1998. Lesser Prairie-Chicken (Tymapanuchus pallidinctus). in The Birds
of North America. No. 364. (F. Gill and A. Poole, editors) The Birds of North
America Inc., Philadelphia, Pennsylvania, USA.
Grimm, V., and I. Storch. 2000. Minimum viable population size of capercaillie Tetrao
urogallus: results from a stochastic model. Wildlife Biology 6: 219-225.
Haukos, D. A., L. M. Smith, and G. S. Broda. 1990. Spring trapping of lesser prairie-
chickens. Journal of Field Ornithology 61: 20-25.
Heppell, S. S., H. Caswell, and L. B. Crowder. 2000. Life histories and elasticity
patterns: perturbation analysis for species with minimal demographic data.
Ecology 81: 654-655.
Hoekman, S. T., L. S. Mills, D. W. Howerter, J. H. Devries, and I. J. Ball. 2002.
Sensitivity analyses of the life cycle of midcontinent mallards. Journal of Wildlife
Management 66: 883-900.
Hulett, G. K., J. R. Tomelleri, and C. O. Hampton. 1988. Vegetation and flora of a
sandsage prairie site in Finney County, southwestern Kansas. Transactions of the
Kansas Academy of Science 91: 83-95.
129
Jamison, B. E. 2000. Lesser prairie-chicken chick survival, adult survival, and habitat
selection and movement of males in fragmented landscapes of southwestern
Kansas. Thesis, Kansas State University, Manhattan, Kansas, USA.
Johnson, K. H., and C. E. Braun. 1998. Viability and conservation of an exploited sage
grouse population. Conservation Biology 13: 77-84.
Knick, S. T., and J. T. Rotenberry. 1995. Landscape characteristics of fragmented
shrubsteppe habitats and breeding passerine birds. Conservation Biology 9: 1059-
1071.
Link, W. A., and P. F. Doherty, Jr. 2002. Scaling sensitivity analysis. Ecology 83: 3299-
3305.
Manly, B. F. 1991. Randomization, bootstrap and Monte Carlo methods in biology. Chapman &
Hall. London, England, UK.
Mills, L. S., D. F. Doak, and M. J. Wisdom. 1999. Reliability of conservation actions
based on elasticity analysis of matrix models. Conservation Biology 13: 815-829.
Nichols, J. D., and J. E. Hines. 2002. Approaches for the direct estimation of λ and
demographic contributions to λ using capture-recapture data. Journal of Applied
Statistics 29: 539-568.
Peterson, M. J., W. T. Grant, and N. J. Silvy. 1998. Simulation of reproductive stages
limiting productivity of the endangered Attwater's prairie-chicken. Ecological
Modelling 111: 283-295.
Peterson, M. J., and N. J. Silvy. 1996. Reproductive stages limiting productivity of the
endangered Attwater's prairie-chicken. Conservation Biology 4: 1264-1276.
Pitman, J. C. 2003. Lesser prairie-chicken nest site selection and nest success, juvenile
gender determination and growth, and juvenile survival and dispersal in
southwestern Kansas. Thesis, Kansas State University, Manhattan, Kansas.
130
Riley, T. Z., and C. A. Davis. 1993. Vegetative characteristics of lesser prairie-chicken
brood foraging sites. Prairie Naturalist 25: 243-248.
Saether, B. E., and O. Bakke. 2000. Avian life history variation in contribution of
demographic traits to the population growth rate. Ecology 81: 642-653.
Schroeder, M. A. and L. A. Robb. 1993. Greater prairie-chicken (Tympanuchus cupido).
in The Birds of North America No. 36. (A. Poole, P. Stettenheim, and F. Gill,
editors) The Birds of North America, Inc., Philadelphia, Pennsylvania, USA.
Steen, H., and K. E. Erikstad. 1996. Sensitivity of willow grouse Lagopus lagopus
population dynamics to variations in demographic parameters. Wildlife Biology
2: 27-35.
Temple, S. A. 1992. Population viability analysis of sharp-tailed grouse metapopulation
in Wisconsin. Wildlife 2001: Populations, 750-758.
Tsai, K., K. H. Pollock, and C. Brownie. 1999. Effects of violation of assumptions for
survival analysis methods in radiotelemetry studies. Journal of Wildlife
Management 63: 1369-1375.
U.S. Department of the Interior, Fish and Wildlife Service. 2002. Endangered and
threatened wildlife and plants; review of species that are candidates and proposed
for listing as endangered or threatened: the lesser prairie-chicken . Federal
Register 67: 40667.
Waddell, B., and B. Hanzlick. 1978. The vanishing sandsage prairie. Kansas Fish and
Game Magazine1-4.
Westemeier, R. L., J. E. Buhnerkempe, and J. D. Brawn. 1998. Effects of flushing
nesting greater prairie-chickens in Illinois. Wilson Bulletin 110: 190-197.
White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from
populations of marked animals. Bird Study 46: 1-31.
131
Wisdom, M. J., and L. S. Mills. 1998. Sensitivity analysis to guide population recovery:
prairie-chicken as an example. Journal of Wildlife Management 61: 302-312.
Wisdom, M. J., L. S. Mills, and D. F. Doak. 2000. Life stage simulation analysis:
estimating vital-rate effects on population growth for conservation. Ecology 83:
628-641.
Withey, J. C., T. D. Bloxton, and J. M. Marzluff . 2001. Effects of tagging and location in
error in wildlife radiotelemetry studies. Pages 45-75 in J. Millspaugh and J.
Marzluff, editors. Radio Tracking and Animal Populations. Academic Press, San
Diego, California, USA.
132
Table 1. Habitat and landscape features for study areas I and II in southwestern Kansas,
1998-2003. Parameter and standard errors (in parentheses) presented when a variable
was not directly measured.
Habitat feature Area I Area II
Fragment size (ha) 5,684 5,660
Sagebrush density (ha-1) 3,611 (689) 4,206 (697)
Sagebrush height (cm)
57 (5.7) 55 (5.9)
Sagebrush diameter (cm) 76 (9.9) 66 (12.5)
Road (km) 154.7 129.2
Powerline (km) 28.4 30.8
Pump-jack (ha-1,000) 10.6 6.9
Buildings (n) 6 1
133
Table 2. Parameters for demographic model of female lesser prairie-chickens.
Parameters describe events occurring within each time step (1 Apr in year t to 1 Apr in
year t + 1).
Parameter Definition
CLUTCH1 Average number of eggs in first nesting attempt.
NEST1y Probability of a yearling’s nest hatching at least one egg.
NEST1a Probability of an adult’s nest hatching at least one egg.
RENEST Probability that a female will renest in the event the first nest
is predated (# of renests / total # of failed nests)
CLUTCH2 Average number of eggs in second nesting attempt, if first
nest fails.
NEST2 Probability of a renest hatching at least one egg.
CHICK Probability that a chick will survive from hatch to 34-d post-
hatch.
PBS Post brood survival (PBS) probability a chick survives from
34-d post-hatch to first spring.
HENy Probability a yearling female will survive from t to t +1.
HENa Probability an adult female will survive from t to t +1.
134
Tabl
e 3.
Par
amet
er e
stim
ates
(θ) o
f vita
l rat
es o
f yea
rling
(Y) a
nd a
dult
(A) l
esse
r pra
irie-
chic
kens
radi
o-m
arke
d in
Fin
ney
Cou
nty,
Kan
sas 1
998-
2002
. Vita
l rat
es a
re d
efin
ed in
Tab
le 2
.
Are
aI
Are
aII
Vita
l rat
e Y
θ ±
SEa
A
θ ±
SE
Y
θ ±
SE
A
θ ±
SE
Pool
ed
θ ±
SE
NES
T 1b
0.30
3 ±
0.08
0 (3
3)c
0.14
0 ±
0.05
3 (4
3)
0.
320 ±
0.09
3 (2
5)
0.41
5 ±
0.07
7 (4
1)
--
NES
T 2
0.14
3 ±
0.16
3 (1
4)
0.12
5 ±
0.08
0 (1
6)
--
REN
EST
0.22
3 ±
0.05
5 (6
0)
0.41
0 ±
0.07
9 (3
9)
--
C
LUTC
H1
----
----
12.1
± 1
.6 (1
51)
CLU
TCH
2--
----
--7.
7 ±
3.0
(29)
CH
ICK
0.
293 ±
0.13
1 (1
2)
0.
209 ±
0.08
0 (2
6)
--
PBSc
----
----
0.53
9 ±
0.08
9 (3
2)
HEN
0.
429 ±
0.11
7 (2
4)
0.30
2 ±
0.08
0 (5
0)
0.
588 ±
0.10
0 (3
3)
0.43
8 ±
0.08
3 (4
8)
--
135
a Clu
tch
size
s wer
e re
porte
d w
ith ±
1 st
anda
rd d
evia
tion,
all
prob
abili
ties w
ith ±
1 st
anda
rd e
rror
, and
sam
ple
size
s are
in p
aren
thes
es.
b All
para
met
er e
stim
ates
exc
ept f
or h
en su
rviv
al w
ere
calc
ulat
ed fr
om fi
eld
data
gat
here
d be
twee
n 19
98 a
nd 2
002.
Ann
ual h
en
surv
ival
was
onl
y es
timat
ed fr
om 2
000
to 2
003.
c Th
e es
timat
es o
f clu
tch
size
s (C
LUTC
H1,
2) a
nd p
ost-b
rood
surv
ival
(PB
S) w
ere
pool
ed a
cros
s are
as a
nd a
ge-c
lass
es d
ue to
con
stan
cy
of ra
tes,
and
sam
ple
size
s, re
spec
tivel
y.
Table 4. Analytic elasticities and variance scaled (arc-sin) sensitivities (VSS) for lower-
level vital rates of matrix.
Area I Area II
Vital ratea Elasticity VSS Elasticity VSS
NEST1y 0.24 0.37 0.10 0.15
NEST1a 0.20 0.49 0.24 0.29
RENEST 0.05 0.11 0.03 0.08
NEST2 0.05 0.09 0.03 0.04
CHICK 0.49 0.76 0.37 0.73
PBS 0.49 0.45 0.37 0.35
HENy 0.23 0.26 0.26 0.22
HENa 0.29 0.43 0.36 0.41
a Vital rates defined in Table 2.
136
Table 5. Results of a life stage simulation analysis (LSA) in which each vital rate was
increased by 10, 20, and 30 % (of its current estimate) and its variability decreased by
10%. The proportion of simulated matrices (n = 1,000) for each vital rate and percent
increase that resulted in λ ≥ 1, is a measure of management effectiveness.
Vital rates (Proportion of 1,000 simulations where λ ≥ 1)
Area / rate increase
NEST CHICK NEST + CHICK
PBS HEN
Area I
10% 0.009 0.007 0.014 0.082 0.010
20% 0.018 0.010 0.032 0.127 0.005
30% 0.027 0.015 0.059 0.180 0.007
Area II
10% 0.026 0.037 0.048 0.062 0.029
20% 0.037 0.066 0.072 0.116 0.057
30% 0.059 0.079 0.133 0.174 0.082
137
Table 6. Results of a modified LSA in which each vital rate was targeted for a set rate of
30 and 40% for nesting and chick survival, and 50 and 60% for post-brood survival and
female survival. Variability was simultaneously decreased by ~20% for each rate. The
proportion of simulated matrices (n = 1,000) for each vital rate and it management goal
that resulted in λ ≥ 1, is a measure of management effectiveness.
Vital rates (Proportion of 1,000 simulations where λ ≥ 1)
Area / targeted rate
NEST CHICK NEST + CHICK
PBS HEN
Area I
30 (20-40)% 0.009 0.000 0.001 NDa ND
40 (30-50)% 0.059 0.005 0.140 ND ND
50 (40-50)% ND ND ND 0.015 0.035
60 (50-70)% ND ND ND 0.036 0.089
Area II
30 (20-40)% 0.004 0.098 0.029 ND ND
40 (30-50)% 0.032 0.286 0.365 ND ND
50 (40-50)% ND ND ND 0.015 0.041
60 (50-70)% ND ND ND 0.077 0.081
a ND = no data recorded for these simulations.
138
Table 7. Ranks of upper-level elasticities from matrix elements for 4 grouse species. A
rank of 1 indicates the most important matrix element.
Ranks of upper-level elasticitya,b
Species F1 F2 F3 P0 P1 P2 Reference
Black grouse 4 3 NDc 2 1 ND Caizergues and Ellison 1997
Greater sage-grouse 6 5 3.5 2 3.5 1 Johnson and Braun 1999
Greater prairie-chicken 3 4 6 1 2 5 Wisdom and Mills 1997
Lesser prairie-chickend 4 2.5 ND ND 2.5 1 This Study
Lesser prairie-chickene 4 2.5 ND ND 2.5 1 This Study
Average ranksf 4.2 3.4 4.8 1.7 2.3 2.0 --
a Ranks of one-half (0.5) indicate a tie in elasticity values, and each was given 0.5 and the
next rank was not used (e.g., a tie of 3 = 3.5, 3.5, 5, 6). b Matrix elements are as follows F1- F3 = fecundity arcs for yearling to 3yr old females,
and P0- P2 = survival (transition) probabilities of juvenile to yearling (P0) through 2 year
old to 3 year olds (P2). c ND = no data available. dArea I in this study. e Area II in this study. f Ranks were averaged to summarize the importance of each rate across species.
139
Table 8. Ranks of lower-level elasticities from matrix elements for 4 grouse species. A
rank of 1 indicates the most important vital rate.
Lower-level elasticity ranksa
Nest
success PBS Adult Reference
Willow ptarmigan 5 4 1 2.5 Steen and Erikstad 1996
Attwaters prairie-chicken 3 1 4 2 Peterson et al. 1998
Capercaillie 3 1 NDe 2 Grimm and Storch (2000)
Lesser prairie-chickenb 3.5 1 3 5 This Study
Lesser prairie-chickenc 5 1 3 3.5 This Study
Average ranksd 3.9 1.6 2.8 3 --
Chick
a Ranks of one-half (0.5) indicate a tie in elasticity values, and each was given 0.5 and the
next rank was not used (e.g., a tie of 3 = 3.5, 3.5, 5, 6). b Area I in this study. c Area II in this study. d Ranks were averaged to summarize the importance of each rate across species. e ND = no data available.
140
Fig. 1. Life-cycle diagram and matrix for a 2 age-class pre-breeding model of female
lesser prairie-chickens. Notations for vital rates are defined in the text.
141
Fig. 2. Results of the life-table response experiment (LTRE) show the contribution of
each vital rate to the difference in λ between Area I (lower half of panel) and Area II
(upper half of panel). This retrospective analysis determined that advantages of NEST1a
and CHICK on Area II and I, respectively, had the largest contribution to population
growth. Notations for vital rates are defined in the text.
143
NEST1y NEST1a RENEST NEST2 CHICK HENy HENa
Con
tribu
tion
to λ
Ι - λΙΙ
= 0
.21
-0.10
-0.05
0.00
0.05
0.10
0.15
144
CHAPTER 5
THE EFFECTS OF LANDSCAPE FEATURES ON LESSER PRAIRIE-CHICKEN
HABITAT USE
Abstract: Suitable habitat for the lesser prairie-chicken (Tympanuchus pallidicinctus) has
been reduced markedly over the past 100 years. The remaining habitat is widely used for
petroleum exploration and extraction, cattle grazing, powerline easements, and the
generation of electricity. Given the tenuous status of the species and the demands on land
use for its remaining habitat, it is imperative that suitable habitat be quantified for impact
assessment and conservation planning. I examined the relationship of several habitat
characteristics and landscape features as they pertained to habitat suitability in
southwestern Kansas. I quantified these characteristics in use and non-use areas as
determined from the presence or absence of prairie chicken locations determined by
radiotelemetry data from 1997 to 1999. I extended the analysis and quantified the
proportion of landscape features and sagebrush density in monthly-ranges of radiomarked
birds 2000 to 2002. The MANOVA results indicated habitat selection by prairie
chickens. Canonical variates analysis resulted in 87 and 13% of the data explained by
canonical variates-1 and -2, respectively. Canonical variate-1 was positively correlated
with sagebrush density, cover, and diameter, but negatively correlated with distance to
wells and structures. This linear combination best described absent-sites. Canonical
variate-2 had the strongest positive correlation with distance to powerline, and best
described use-sites. The average odds of a powerline occurring in a non-range were 3.22
times more likely than in a monthly-range. Monte Carlo simulations of distances to
145
structures indicated that the nearest 40% of lesser prairie-chicken centers of use were
farther from anthropogenic features than would be expected at random.
INTRODUCTION
Suitable habitat for the lesser prairie-chicken (Tympanuchus pallidicinctus) has
been markedly reduced over the past 100 years. What remains is highly fragmented
throughout the species’ range (Giesen 1998). The remaining habitat is widely used for
petroleum exploration and extraction, cattle grazing, powerline easements, and the
generation of electricity (Jamison et al. 2001a). The cumulative loss of habitat and
declining population trends led to the “warranted but precluded” threatened listing under
the Endangered Species Act (USFWS 2002). The lesser prairie-chicken’s dependency on
native rangeland has been well studied, at the microscale (Jones 1963, Crawford and
Bolen 1974, Riley et al. 1992, Jamison et al. 2001b), and the macroscale (Jamison 2000,
Woodward et al. 2001, Fuhlendorf et al. 2002). Despite these large-scale contractions of
habitat and thorough examination of microhabitat variables, little is known about the
specific parameters of a suitable habitat patch. Given the tenuous status of the species
and the demands on land use for its remaining habitat, it is imperative that suitable habitat
be quantified for conservation assessment and planning.
Jamison (2000) examined habitat use of male lesser prairie-chickens as it
pertained to selection ratios of sand sagebrush (Artemisia filifolia), cropland, and
miscellaneous grasslands in southwestern Kansas. Generally, lesser prairie-chickens
were tied to sagebrush throughout the year, but exhibited higher selection ratios for this
cover-type during the summer months (Jamison 2000). Despite the apparent importance
of sagebrush, there were several areas within the prairie fragment (~5,000 ha) in which
146
radiomarked birds were not located, and unmarked individuals were not flushed or
observed in these areas (B. E. Jamison, unpublished data). This lack of habitat use
provided an opportunity to quantify both habitat variables and landscape features as they
pertain to suitability of lesser prairie-chicken habitat.
The objective of this paper is to examine what suite of habitat characteristics and
landscape features determines habitat suitability in southwestern Kansas. I used
Jamison’s (2000) radiotelemetry location data of male and female lesser prairie-chickens
to quantify these characteristics in use and non-use sites as determined from the presence
or absence of prairie chicken locations. I extended the analysis in 2000-2002 and
quantified the proportion of landscape features and sagebrush density in monthly-ranges
of radiomarked females.
METHODS
Study area
The study region was comprised of 2 fragments (~5000 ha each) of native
sandsage prairie near Garden City, Finney County, Kansas (37° 52′ N, 100° 59′ W).
Work began on Area I (southwest of Garden City) in 1997. In 2000, trapping and
monitoring efforts were expanded to include Area II (southeast of Garden City). Prior to
1970, these 2 areas were part of a contiguous tract of native sandsage prairie. The
development of center pivot irrigation left these areas as two fragments with about 19 km
of non-habitat between patch centroids (Waddell and Hanzlick 1978). Shrub and grass
vegetation in the prairie fragments was comprised of sandsage, yucca (Yucca spp.),
sandreed grasses (Calamovilfa spp.), bluestem grasses, sand dropseed (Sporobolus
cryptandrus), and sand lovegrass (Eragrostis trichodes). Primary forb species in the
147
region included ragweed (Ambrosia spp.), sunflower (Helianthus spp.), and Russian
thistle (Salsola iberica) (Hullett et al. 1988, Jamison 2000).
Field methods
Lesser prairie-chickens were captured on leks using walk-in funnel traps during
March and April 1997-2002 (Haukos et al. 1991). Each captured bird was fitted with a
lithium battery powered transmitter with a mass < 12-g. During Phase I (1997–1999)
males and females were tracked daily from April to September 1997, and April to April
1998–2000. Females were tracked daily from April to April during Phase II (2000-
2003). A truck-mounted null-peak twin-Yagi telemetry system was used to triangulate
individuals remotely.
Retrospective analysis (Phase I)
Defining use.–I quantified both micro- and macrohabitat variables that occurred in
use and non-use areas as determined from Jamison’s (2000) study from 1997 to 1999.
Thus, I sampled vegetation in the areas defined below in 2000 and 2001. Use areas were
defined from a 95% fixed kernel home range (Worton 1989) of all lesser prairie-chicken
locations (n = 21,363) that intersected the study area (Jamison 2000). Areas not
intersected by this polygon were defined as non-use areas. I defined 2 types of non-use
sites: 1) areas adjacent to use sites but where radiomarked birds were never located
(hereafter non-use sites), and 2) an area north of the Sunflower Electric Power plant
(Area I southwest of Garden City, KS) which was not known to have breeding
populations of prairie chickens, and where radiomarked birds were never located
(hereafter absent sites). Although I cannot make absolute statements regarding the
148
avoidance of these areas by lesser prairie-chickens, the data recorded over 6 years never
documented radiomarked or unmarked birds in absent and non-use areas.
Sampling.–I generated random points (UTM coordinates) within use-, non-use,
and absent areas in ARCVIEW 3.1 (ESRI 1998). Random points were located in the
field using a handheld global positioning system (GPS) unit. At each random point, I
placed a 1.5 × 0.5 m quadrat and estimated the percent cover of shrub, grass, forb, and
litter. Shrub density ha-1 was estimated using a modification of the point center quarter
method (Cottam and Curtis 1956), where if the nearest shrub was > 15 m from the
random point that quadrant was censored, and density was estimated based upon both the
number of distances measured and the number censored (T. Loughin, Kansas State
University, unpublished data). This modification was necessary to increase the efficiency
of data collection in the field. Distances to the nearest road, powerline, natural gas or oil
well-head (hereafter wells), and structure (e.g., dwellings, compressor station, ash-pile or
powerplant) from each random point were calculated in a geographic information system
(GIS).
Home range estimation and GIS (Phase II)
For the location data from 2000 to 2002 the analyses were extended and
quantified the proportion of landscape features and sagebrush density in monthly-ranges
of radiomarked females. Azimuths from fixed listening stations were entered into Locate
II triangulation software (Nams 2002), and locations of prairie chickens were estimated
using Lenth maximum likelihood estimators. Location data were imported into a GIS
database of the study area and monthly ranges were estimated using a 95% fixed kernel
(Worton 1989) in ARCVIEW 3.1. I limited my sample to those birds that had ≥20
149
locations per month. While this reduced the number of observational units, it ensured
that observed range sizes were likely representative of the area covered in a month
relative to birds with fewer locations. This also limited the season for which such an
analysis could be conducted as the season progressed the number of individuals tracked
decreased and premature battery failure also contributed to reductions in sample size.
Thus, I chose to focus the analysis on times of the year (April to September) when
samples were relatively large and coincided with a peak usage of sagebrush habitats
(Jamison 2000). All female locations that occurred on nest sites were excluded as the
relationship between nesting and these variables has already been examined (Pitman
2003).
Locations of wells, roads, powerlines, and structures were digitized into the GIS.
Monthly ranges for each bird were computed separately for year and month (e.g., Apr
2000) to control for yearly and biological variation in monthly habitat use. The ranges
for all birds in a given year and month were combined into a single overall “use range”
which was overlaid onto the study area. Each year and month layer (hereafter a
monthly-range) was used as a sampling frame to calculate the number of wells, roads,
powerlines and structures falling within and outside of (hereafter non-ranges) prairie
chicken monthly-ranges. Wells could be counted directly, however, roads and
powerlines could not. Thus, following the procedure of Marcum and Loftsgaarden
(1980) 244,000 random points were generated within each of the study area polygons and
the number of points that intersected roads and powerlines in use and non-use areas
tabulated. This resulted in count-data that could be input into contingency table or
Poisson rate regression analyses.
150
Analysis
Retrospective analysis of use.—A fixed model MANOVA (PROC GLM: SAS)
was used to examine habitat use for a community of variables (Johnson 1998). This
model allowed me to compare the effect of habitat characteristics and landscape features
on use simultaneously. I used MANOVA as a conservative approach to multiple
comparisons, and to examine the dimensionality of the alternative hypothesis using
canonical variates analysis (Johnson 1998). Thus, I reported the means and standard
errors of habitat and landscape variables, and their resulting P-values from multiple
comparisons. I examined the possibility of explaining these sites in a reduced parameter
space (i.e. the dimensionality of the alternative hypothesis). The dimensionality of Ho =
3, and Ha < 3 in this study. If the population means lie in a subspace (data reduction)
then linear combinations (canonical variates) can be used to explain the relationship
between use, non-use, and absent sites. I tested the differences in mean canonical
variates using differences of means and 95% confidence limits around the differences,
and the degree of overlap of 95% ellipses in a 2-dimensional space (canonical variate-1 ×
canonical variate-2). The approximate 95% ellipses were estimated as
nk
2,αχ
,
where χ2 = chi-square value at α = 0.05, k = number of dimensions, n = number of
populations, and if k = 2 then the ellipses are circles.
Six habitat variables were included in analyzing use, non-use, and absent sites:
percent cover estimates for sagebrush, grass, forb, and litter; shrub density, and shrub
diameter. Distances from random points (within the areas defined above) to the nearest
151
structure, well, powerline, and road (included both paved and unimproved trails) were
also included in the analysis.
Monthly-ranges and use.—In the following analyses it is important to recognize
that monthly usage included repeated measures on individual birds. This limits my
inference and may increase my Type I error rate if certain individuals had a propensity to
remain farther from these features than the population mean. However, only 60% of the
birds had consistent repeated measures through the sampling period, making a repeated
measures design impractical. Thus, I treated monthly-range as the observational unit and
blocked on month. Because the goal of the analysis is not to differentiate the variation in
usage across months but to examine general patterns of use as it pertains to these
landscape features, I assessed the effects on use and controlled for the monthly
variability by blocking on month.
I used 2 × 2 × K contingency tables and Poisson rate regression to examine the
relationships between monthly-ranges and the proportion of each area occupied by
powerlines, roads, and wells controlling for months there were 6 blocks in 2000 and
2001, and 4 blocks in 2002. Contingency tables were used to examine the proportion of
roads and powerlines present in monthly-ranges compared to non-range areas. The
Breslow-Day test for homogeneity of odds ratios was used to test for a common odds
ratios across months (Agresti 1996). Poisson rate regression was used to model the
number of wells present in monthly-ranges as a function of use month and year.
Backward selection was used to find the most parsimonious model that fit well, and
model fit was assessed by examining the scale parameter (deviance / df) and residual
plots.
152
Sagebrush density was estimated for monthly- and non-ranges of lesser prairie-
chickens using the modification of the point-center quarter (PCQ) method described
above. Density was estimated from a database of >3,000 points where PCQ
measurements were taken. I tested for the effect of use on differences in density between
ranges using a mixed model ANOVA, blocking on month.
Permutations of structure distance.—Because structures were not randomly or
uniformly located on the landscape in relation to monthly-ranges, I used a modified
Monte Carlo simulation (Manly 1997) to test whether the centroids of monthly-ranges
were farther from structures than would be expected at random. If the structures had no
impact on the birds' monthly-ranges, then one would expect the distances between these
centroids and the structures to follow the same distribution as distances to randomly-
placed points. If, on the other hand, the birds demonstrated some avoidance of these
structures, as hypothesized, then the centroids closest to these structures should tend to be
farther away than the random points. I therefore compared the 5th, 10th, 20th, 30th and 40th
percentiles of the observed distances to the respective distributions of the correcponding
percentiles from randomly-placed points as follows: 1) random resamples of 81 distances
were taken from the distribution of 100,000 random points and distances; 2) this was
repeated 1,000 times (thus generating 1,000 resamples of 81 distances); 3) the 5th, 10th,
20th, 30th and 40th percentiles for each resample were calculated; 4) an estimated sampling
distribution of each percentile was created based on the 1000 calculated percentiles; and
5) the observed percentile distances were compared to the the upper tail of their
respective sampling distribution. If the observed percentile distance was too extreme in
the upper tail (i.e. in the upper 5% of the distribution), the Ho (structures do not affect
153
monthly-range placements) was rejected. Support for the Ha, that structures have a
negative impact on habitat use was then examined. The observed percentiles and the
expected (mean) value of that percentile under randomness were compared.
RESULTS
Use and non-use
A total of 281 lesser prairie-chickens were captured and 160 (male = 76, female =
84) were fitted with transmitters. This resulted in 21,373 daily locations from 1997 to
1999. Vegetation sampling points in use (n = 44), non-use (n = 38), and absent (n = 46)
areas were also used to calculate distances to landscape features. The MANOVA results
indicated habitat selection by prairie chickens (Wilks Λ = 0.352, F20,222 = 7.58, P <
0.0001) with large numbers of multiple comparisons with effect sizes 1- to 2-fold
differences between means (Table 1). The dimensionality of Ha = 2 (canonical variate-
1, F20,222 = 7.58, P < 0.0001; canonical variate-2, F9,112 = 2.48, P = 0.013) and canonical
variates-1 and -2 (CAN-1 and CAN-2) explained 87 and 13% of the data, respectively.
Canonical variate-1 was positively correlated with sagebrush density, cover, and diameter
(Table 1), but negatively correlated with distance to wells and structures; collectively this
describes absent-sites. Canonical variate-2 had the strongest positive correlation with
distance to powerline, and use-sites (Table 2). Thus, lesser prairie-chicken habitat use
was generalized by the linear combinations, CAN-1 and CAN-2, of sagebrush stand
characteristics and well proximity, and distance to powerlines, respectively. Effect sizes
and 95% confidence intervals of the mean CAN-1 scores (Table 2) for use, non-use, and
absent areas indicated that differences in shrub habitat were similar between use and non-
use sites ( usenonuse xx −− = 0.365, 95% CI: -0.070, 0.800), but differed markedly from
154
absent sites ( usenonabsent xx −− = 2.621, 95% CI: 2.183, 3.059; useabsent xx − = 2.256, 95% CI:
1.816, 2.696). This was supported by sagebrush density estimates obtained from
calculations of all sampling points within each area. Densities were: absent = 7, 205 SE
= 573; non-use = 2,553 SE = 201; and use sites = 3,144 SE = 249 ha-1. Conversely,
CAN-2 scores revealed (Table 2) greater similarity between non-use and absent sites in
terms of distance to landscape features ( usenonabsent xx −− = 0.403, 95% CI: -0.034, 0.841)
than between either of these areas, and use sites ( usenonuse xx −− = 0.661, 95% CI: 0.221,
1.101 absentuse xx − = 1.064, 95% CI: 0.629, 1.499). Examination of the 2-dimensional
plot and 95% confidence ellipses indicated complete separation between use, non-use,
and absent sites (Fig. 1), along the sagebrush and landscape feature gradients.
Monthly-ranges
A total of 147 females were captured and fitted with radio transmitters from 2000
to 2002; only 95 birds provided enough data to be included in the analysis of 283
monthly-ranges (Table 3). The Breslow-Day test for the odds of roads occurring in
monthly-ranges indicated that the odds ratios were significantly different across months
in 2 of 3 years (2000 χ2 = 28.62, df = 5, P < 0.0001: 2001; χ2 = 29.07, df = 5, P <
0.0001: 2002; χ2 = 80.58, df = 3, P < 0.0001) for each year (Fig. 2). The odds of roads
in a monthly-range were generally about 11% less than that of non-range sites. The
Breslow-Day test for the odds of powerlines occurring in monthly-ranges indicated that
the odds ratios varied throughout the sampling period (2000 χ2 = 9.61, df = 5, P =
0.087: 2001; χ2 = 403.73, df = 5, P < 0.0001: 2002; χ2 = 21.52, df = 3, P < 0.0001).
The average odds of a powerline occurring in a monthly-range were 3 times (n = 16, SD
= 0.28) less likely than in a non-range (Fig. 3). However, the odds increased in 2002 as
155
a number of birds had home ranges that overlapped with a newly constructed powerline.
Sagebrush density was higher ( usenonuse xx −− = 817 ha-1; F1,18.6= 63.28, P < 0.0001) in
monthly-ranges ( x = 3,743; SE = 62.7 ha-1) than areas outside of monthly-ranges ( x =
2,996; SE = 62.7 ha-1)
Backward selection of Poisson rate regression models of counts of wells as a
function of prairie chicken use indicated that a model containing year × month and month
× use interaction and all main effects was the best fit (deviance / df = 0.448). Estimates
of the mean differences in number of wells per monthly-range varied by month (Fig. 4)
with the months of April, July, and September having slightly fewer wells ha-1,000 than
non-range sites.
Monte Carlo simulations of distances to structures indicated that average
monthly-centroids (O) were farther than expected at random (E) for the 5th (E = 967 m;
O = 1,228 m; P < 0.001), 10th (E = 1,241 m; O = 1,403 m; P < 0.001), 20th (E = 1,615
m; O = 1,654 m; P < 0.001), 30th (E = 1,759 m; O = 1,785 m; P < 0.05), and 40th
percentiles (E = 1,998 m; O = 2,006 m; P < 0.05).
DISCUSSION
This study is the first to document meso-scale factors that likely contribute to
reduced suitability of lesser prairie-chicken habitat. The landscape features, proportion
of an area occupied by powerlines and proximity to human structures, clearly reduced use
of otherwise suitable habitat. This was evidenced by the greater distance between prairie
chicken use sites and powerlines (1–2 fold difference), and the low odds ratios of
poweline occurrence. Monte Carlo simulations indicated that the nearest 40% of lesser
prairie-chicken centers of use were farther from anthropogenic features than would be
156
expected at random. However, the relationship of habitat usage as it pertains to
sagebrush density and cover was less clear. Generally, it would appear that sagebrush
densities in use areas were slightly higher on average than in non-use areas, but absent
sites had larger shrubs and at higher estimated densities. This suggests that the habitat
may otherwise be suitable, and further work is needed to discern these relationships.
The other factors that were important to the absent area were the proximity of wells and
structures as they were negatively related with CAN-1. Based on their mean distance
values of 759 and 323 m, wells and structures were significantly closer to sampling
points, respectively, on the absent site than either use or non-use sites. However, non-use
sites were 336 m closer to strucures than use sites. Combined, these data suggest that
lesser prairie-chickens used areas with less anthropogenic disturbance.
Roads seemed to have little impact on prairie chicken habitat use, however the
roads included here were of various quality: from paved 2-lane highways to unimproved
2-tracks through the sandhills. Pitman (2003) found that nesting female lesser prairie-
chickens placed nest sites farther from paved roads than would be expected at random.
Given the relatively large ranges ( x = 326 ha or 3.2 km2) of female prairie chickens in
this study it is not surprising that the proportion of roads was not significantly different
across area types. Oyler-McCance (1999) modeled Gunnison sage-grouse (Centrocercus
minimus) patch occupancy and found that the probability of patch occupancy was
positively correlated with distance from a paved road. Patch occupancy was best
described by landscape scale features when compared to models that included
microhabitat parameters (Oyler-McCance 1999). The potential for landscape features to
157
reduce the suitability of microhabitat variables may be more widespread, as was indicated
in my study.
Previous work indicated that changes in habitat composition at large scales (7,200
ha) explained most of the variability in declining lesser prairie-chicken populations
(Woodward et al. 2001, Fuhlendorf et al. 2002). Fuhlendorf et al. (2002) did find that
increased edge density was a significant factor in declining populations at the 3 smallest
scales (405, 952, and 1810 ha) in their study, and they suggested that such changes have
the greatest impacts on breeding activities. If such increased edge density was correlated
to powerline, well, or structure density (in this study) and decreased reproductive output,
then it could explain in part why the absent and non-use sites were less suitable for lesser
prairie-chicken occupancy. Woodward et al. (2001) reported that population indices
remained stable in landscapes where shrub cover was lost at a slower rate. While such
losses have not occurred on these study areas in 15–20 years, the internal fragmentation
and habitat loss (i.e., road and powerline construction) likely are contributing to the
unsuitable portions of the study area.
Earlier work (Crawford and Bolen 1976, Taylor and Guthery 1980, Jamison
2000) described habitat use and suitability as it pertained primarily to the proportion of
agriculture in a landscape. Crawford and Bolen (1976) suggested that when landscapes
reach <63% native rangeland they are unsuitable habitat. Taylor and Guthery (1980)
suggested that a minimum of 3.2 km2 is needed to maintain prairie chickens. Although
important, these studies failed to describe the necessary cover and other extrinsic factors
within a fragment that make it suitable for prairie chickens.
158
MANAGEMENT IMPLICATIONS
Future impact assessments should consider the construction of new anthropogenic
features as a potential detriment to habitat suitability for lesser prairie-chickens.
Conservation planning and habitat inventory should carefully consider the proximity and
density of anthropogenic features when prioritizing habitat patches for conservation
action or mitigation. Further work is needed to clarify the potential interaction between
sagebrush stand characteristics (density and plant morphology) and landscape features as
it pertains to habitat suitability. Future research should examine how the quantities (or
proximity) of anthropogenic features affect the probability of patch occupancy across
replicate sites of various sizes and levels of disturbance.
LITERATURE CITED
Agresti, A. 1996. An introduction to categorical data analysis. Wiley & Sons Inc.. New
York, New York, USA.
Crawford, J. A., and E. G. Bolen. 1976a. Effects of land use on lesser prairie chickens in
Texas. Journal of Wildlife Management 40:96-104.
Cottam, G., and J. T. Curtis. 1956. The use of distance measures in phytosociological
sampling. Ecology 37: 451-460.
ESRI. 1998. Arc View GIS 3.1, Online users manual. Environmental Systems Research Institute,
Redlands, California, USA.
Fuhlendorf, S. D., A. J. Woodward, D. M. Leslie, and J. Shackford. 2002. Multi-scale effects of
habitat loss and fragmentation on lesser prairie-chicken populations in US Southern Great
Plains. Landscape Ecology 17: 617-628.
Giesen, K. M. 1998. Lesser Prairie-Chicken (Tymapanuchus pallidinctus). in The Birds
159
of North America. No. 364. (F. Gill and A. Poole, editors) The Birds of North
America Inc., Philadelphia, Pennsylvania, USA.
Haukos, D. A., L. M. Smith, and G. S. Broda. 1990. Spring trapping of lesser prairie-
chickens. Journal of Field Ornithology 61: 20-25.
Hulett, G. K., J. R. Tomelleri, and C. O. Hampton. 1988. Vegetation and flora of a
sandsage prairie site in Finney County, southwestern Kansas. Transactions of the
Kansas Academy of Science 91: 83-95.
Jamison, B. E. 2000. Lesser prairie-chicken chick survival, adult survival, and habitat
selection and movement of males in fragmented landscapes of southwestern
Kansas. Thesis, Kansas State University, Manhattan, Kansas, USA.
Jamison, B. E., J. A. Dechant, D. H. Johnson, L. D. Igl, C. M. Goldade, and B. R. Euliss .
2002a. Effects of management practices on grassland birds: lesser prairie-chicken.
U.S. Geological Survey, Northern Prairie Wildlife Research Center. Jamestown,
North Dakota, USA.
Jamison, B. E., R. J. Robel, J. S. Pontius, and R. D. Applegate. 2002b. Invertebrate
biomass: associations with lesser-prairie chicken habitat use and sand sagebrush
density in southwestern Kansas. Wildlife Society Bulletin 30: 517-526.
Johnson, D. E. 1998. Applied multivariate methods for data analysts. Duxbury Press.
Pacific Grove, California, USA.
Jones, R. E. 1963. Identification and analysis of lesser and greater prairie chicken habitat.
Journal of Wildlife Management 27:757-778.
Manly, B. F. 1997. Randomization, bootstrap and Monte Carlo methods in biology. Chapman &
Hall. London, England, UK.
Marcum, C. L., and D. O. Loftsgaarden. 1980. A non-mapping technique for studying
habitat preferences. Journal of Wildlife Management 44: 963-968.
160
Nams, V. 2000. Locate II: user’s guide. Pacer Computer Software, Truro, Nova Scotia,
Canada.
Oyler-McCance, S. J. 1999. Genetic and habitat factors underlying conservation
strategies for Gunnison sage-grouse. Dissertation, Colorado State University, Fort
Collins, Colorado, USA.
Pitman, J. C. 2003. Lesser prairie-chicken nest site selection and nest success, juvenile
gender determination and growth, and juvenile survival and dispersal in
southwestern Kansas. Thesis, Kansas State University, Manhattan, Kansas.
Riley, T. Z., C. A. Davis, M. Oritz, and M. J. Wisdom. 1992. Vegetative characteristics
of successful and unsuccessful nests of lesser prairie-chicken. Journal of Wildlife
Management 56: 383-387.
Taylor, M. A., and F. S. Guthery. 1980. Fall-winter movements, ranges, and habitat use
of lesser prairie-chickens. Journal of Wildlife Management 44: 521-524.
U.S. Department of the Interior, Fish and Wildlife Service. 2002. Endangered and
threatened wildlife and plants; review of species that are candidates that are
proposed for listing as endangered or threatened: the lesser prairie-chicken.
Federal Register 67: 40667.
Waddell, B., and B. Hanzlick. 1978. The vanishing sandsage prairie. Kansas Fish and
Game Magazine 1-4.
Woodward, A. J., S. D. Fuhlendorf, D. M. Leslie, and J. Shackford. 2001. Influence of
landscape composition and change on lesser prairie-chicken (Tympanuchus
pallidinctus) populations. American Midland Naturalist 145: 261-274.
Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in home-
range studies. Ecology 70:164-168.
161
Tabl
e 1.
Par
amet
er e
stim
ates
of h
abita
t and
land
scap
e fe
atur
es fo
r les
ser p
rairi
e-ch
icke
n us
e, n
on-u
se a
nd a
bsen
t site
s. M
eans
and
stan
dard
err
ors a
re p
rese
nted
for m
easu
red
varia
bles
alo
ng w
ith th
e st
anda
rdiz
ed c
anon
ical
scor
es (C
AN
-1 a
nd C
AN
-2) f
or e
ach
varia
ble.
Col
umns
with
diff
eren
t let
ters
indi
cate
a si
gnfic
ant d
iffer
ence
(P <
0.0
5).
Are
a ty
pe
Stan
dard
ized
can
onic
al sc
ores
Var
iabl
e
Use
Non
-use
Abs
ent
C
AN
-1C
AN
-2Sh
rub
cove
r (%
) 15
.3 ±
3.1
A
13.6
± 3
.1A
35
.0 ±
3.1
B
0.42
8 -0
.559
Forb
cov
er (%
) 9.
7 ±
1.4A
7.
7 ±
1.4A
8.
9 ±
1.5A
0.44
10.
015
Gra
ss c
over
(%)
13.4
± 1
.7A
14
.3 ±
1.6
A
14.1
± 1
.7A
-0
.360
-0.5
21
Litte
r cov
er (%
) 29
.9 ±
2.4
A
38.6
± 2
.4B
16
.4 ±
2.5
C
0.24
80.
028
Shru
b de
nsity
(m2 )
2.5 ±
0.2A
2.
0 ±
0.2A
4.
4 ±
0.2B
0.
870
0.29
0
Shru
b di
am (c
m)
71.0
± 3
.2A
67
.7 ±
3.1
A
80.1
± 3
.3B
0.
217
-0.2
13
Dis
tanc
es (m
)
W
ells
43
5.0 ±
31.2
A
446.
0 ±
30.8
A
323.
2 ±
31.6
B
-0.3
820.
104
Roa
ds
193.
2 ±
18.7
A
178.
8 ±
18.6
A
184.
4± 1
9.0A
-0
.071
0.11
4
Stru
ctur
e 1,
397 ±
106A
1,
061 ±
105B
75
9 ±
108C
-0
.332
-0.0
40
Pow
erlin
es
1,10
6 ±
81A
66
6 ±
80B
70
5 ±
82B
-0
.060
1.02
4
162
Table 2. Canonical scores and sagebrush density plants ha-1 ( x ± SE) of
lesser prairie-chicken use area type. Differences in usage area type are defined in the text.
Area type
Variate/ sagebrush Use Non-use Absent
CAN-1 – 0.561 ± 0.158 – 0.926 ± 0.154 1.694 ± 0.158
CAN-2 0.591 ± 0.158 – 0.472 ± 0.154 – 0.069 ± 0.158
Sagebrush density ha-1 3,144 ± 249 2,553 ± 201
7,205 ± 573
163
Table 3. Count data of landscape features for contingency table and Poisson rate
regression modeling of lesser prairie-chicken monthly-ranges.
Year Apr May Jun Jul Aug Sep
2000 No. ranges 30 8 21 23 17 17
Avg. ha 553 667 273 170 148 180
Points in ranges 111,493 72,929 5,258 33,076 35,084 3,561
Roads (points) 2,597 1,575 1,127 617 829 752
Powerlines (points) 150 58 37 26 38 27
No. wells 52 44 21 10 13 8
2001 No. ranges 15 9 18 24 25 23
Avg. ha 805 1045 396 182 119 151
Points in ranges 97,064 107,238 78,575 42,787 41,272 34,033
Roads (points) 2233 2522 2025 1024 951 763
Powerlines (points) 144 213 398 54 38 60
No. wells 44 58 39 17 19 13
2002 No. ranges 11 8 16 16 ND
ND
Avg. ha 584 231 306 135 ND ND
Points in ranges 61,526 27,107 64,288 30,491 ND ND
Roads (points) 1,729 600 1,420 672 ND ND
Powerlines (points) 540 241 518 186 ND ND
No. wells 24 12 34 10 ND ND
164
Fig 1. Mean canonical variates (CAN-1 and CAN-2), SEs (in both x,y) and their
respective 95 % confidence circles for lesser prairie-chicken use, non-use and absent
sites. CAN-1 was best explained by high sagebrush densities and large shrubs. Distances
to structures and wells also explained CAN-1, but were negatively related. This revealed
that that absent sites were in closer proximity to these anthropogenic features than other
sites. CAN-2 was best explained by distance to powerlines. Although these sites
separated in a 2-dimensional space, use and non-use did not differ in regards to their
mean value of CAN-1, but both differed from the mean CAN-1 score of absent sites.
Alternatively, non-use and absent sites were similar with respect to CAN-2 but differed
from use sites, suggesting that distance to powerlines was important in determining
suitability of lesser prairie-chicken habitat.
165
Fig. 2. Odds ratios (95% confidence limits) for roads occurring in lesser prairie-chicken
monthly-ranges in 2000 (A), 2001 (B), and 2002 (C). The dashed line indicates odds of
1 and confidence limits intersecting this line indicate odds not different than expected by
chance.
167
2000
Apr May Jun Jul Aug Sep
Odd
s of
road
s in
mon
thly
-rang
e
0.6
0.7
0.8
0.9
1.0
1.1
2001
Apr May Jun Jul Aug Sep0.80
0.85
0.90
0.95
1.00
1.05
1.10
1.15
2002
Apr May Jun Jul0.7
0.8
0.9
1.0
1.1
1.2
1.3
(A) (B)
(C)
168
Fig. 3. Odds ratios (95% confidence limits) for powerlines occurring in lesser prairie-
chicken monthly-ranges in 2000 (A), 2001 (B), and 2002 (C). The dashed line indicates
odds of 1 and confidence limits intersecting this line indicate odds not different than
expected by chance.
169
2000
Apr May Jun Jul Aug Sep0.00
0.05
0.10
0.15
1.00
2001
Apr May Jun Jul Aug Sep0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
2002
Apr May Jun Jul0.4
0.5
0.6
0.7
0.8
0.9
1.0
Odd
s of
pow
erlin
es in
mon
thly
-rang
e
(A) (B)
(C)
170
Fig. 4. Differences in the mean number (95% confidence limits) of wells ha-1,000 between
monthly- and non-ranges as determined from Poisson rate regression. The dashed line
indicates a difference of 0 and confidence limits intersecting this line indicate a observed
values did not differ from those expected by chance.
171
Apr May Jun Jul Aug Sep
Diff
eren
ce in
no.
of w
ells
ha-1
,000
(95%
CL)
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
172
OVERALL STUDY SUMMARY
Introduction
The lesser prairie-chicken (Tympanuchus pallidicinctus) has sustained marked
reductions in suitable habitat over the past 100 years. The remaining habitat is widely
used for petroleum exploration and extraction, cattle grazing, powerline easements, and
the generation of electricity. It is estimated that over 90% of the range has been lost since
the turn of the 20th century and the range wide population has paralleled these losses. In
1995, the lesser prairie-chicken was petitioned to be listed as threatened under the
Endangered Species Act. Given the status of this sensitive species it is critical to
understand the dynamics of local and regional populations and possible management
scenarios that would most effectively benefit the lesser prairie-chicken.
This study is the culmination of two Federal Aid in Wildlife Restoration projects,
W-47-R and W-53-R. Field work for W-47-R began in spring 1997 and was completed
in November 1999, and W-53-R was initiated in December 1999 and concluded in March
2003. Project W-47-R was located on a 5,700-ha fragment of sand sagebrush prairie
(Area I) south of the Arkansas River (~ 4 km south of Holcomb, Finney County, Kansas).
Project W-53-R added second study site (Area II) of 5,400 ha. The center of Area II was
~20 km east of Area I. Research was conducted simultaneously on both sites from spring
2000 through spring 2003. Primary research objectives were to gather information on
specific demographic parameters and examine population viability and sentivity to
changes in the parameters.
173
A total of 755 lesser prairie-chickens were captured and banded from 1997 to
2002, and 136 male (46 yearling, 66 adult, 24 age undetermined), and 227 female (87
yearling, 117 adult, and 23 age undetermined) were fitted with radio transmitters.
Radiomarked birds were tracked daily using vehicle mounted antennae, resulting in
39,969 daily bird locations. These data were used to determine nest success, chick
survival, adult survivorship, habitat usage, and the resulting parameter estimates were
used as inputs to the demographic model.
Age-specific survival of males
Robust estimates of annual survival are useful for two reasons: understanding
management efforts and basic science. Survival is one of several demographic rates that
can effect the fluctuations in population numbers of grouse and the occurrence of age-
specific survival and reproductive rates may covary with the type of mating system in
grouse. Three hundred and seventy-six male prairie-chickens (173 yearlings, 203 adults)
were captured from 1998-2002, and 150 males (78 yearlings, 72 adults) were recaptured
at least once. Analyses of live mark-recapture data were used to estimate survival. Local
survival rates (φ) of male lesser prairie-chickens were highest for yearling (φ1 = 0.615) >
adult (φ1 = 0.485) > older adults (φ2 = 0.347). Twenty percent of recaptured yearlings
switched leks in their second year, and were 2.5 (odds ratio) times as likely than adults to
do so (8%). Each age-class was equally likely (~17.5 %) to move between leks within a
breeding season. Four and 15 % of yearlings and adults were recaptured > 3 times,
respectively. This suggests that adults had a greater propensity to attend leks and
presumably try to obtain copulations than yearlings. The overall estimates of lesser
174
prairie-chicken survival were slightly elevated when compared to that of other banding
studies of prairie grouse.
Effects of radiomarking on lesser prairie-chicken survival
Radiotelemetry has provided wildlife biologists a tool to estimate survival where
fate of each individual is likely known. While these analyses of these data can result in
highly accurate estimates, five assumptions must be met. Two of the assumptions,
random right-censoring and no-transmitter effect on survival, are often difficult to assess
in studies. Male lesser prairie-chickens in this study were not measurably impacted by
radiomarking, as their survival rates were greater than or equal to those of banded birds.
The model best supported by the data, Sc, pgroup+t, rg, Fc, indicated that survival was best
modeled as constant (Ŝc = 0.731, SE = 0.072) across radiomarked and banded birds.
Signal loss occurred throughout the monitoring period and appeared to be independent of
periods of high mortality. Eight of 16 (50%) right-censored birds were subsequently
recaptured, which was similar to the recapture rates for known-fate birds (55.8%),
suggesting that right-censored birds had similar survival rates to that of known-fate
individuals.
Survival of male and female lesser prairie-chickens
In total, 136 male (46 yearling, 66 adult, 24 age undetermined), and 227 female
(87 yearling, 117 adult, and 23 age undetermined) lesser prairie-chickens were captured
and fitted with radio transmitters. Year to year variation in summer survival was evident
and seemed to be highly related to mortality rates of incubating females on nests and
>30% of all female mortality was associated with the nesting period. Summer survival
rates (Apr-Nov) and apparent nest success from 5 years revealed the strength of this
175
relationship. Gender-specific survival was not well supported by the data, as the overall
survival rates of males and females were similar in the summer (Smale = 0.69 ; Sfemale =
0.67) and seasonal (Apr-Mar) analyses (Smale = 0.39; Sfemale = 0.43). However, the
timing of survival over 12-months suggested that seasonal patterns in male survivorship
do not match the temporal pattern seen in females.
Age-specific patterns in female survival were evident in both analyses with
yearling birds (Ssummer = 0.75; Sseasonal = 0.52), surviving at a higher rate than adults
(Ssummer = 0.69, Sseasonal = 0.37). One explanation for this difference is the cost of
reproduction, as it may result in reduced survival rates of adults, if they had higher
reproductive output, but, reproductive parameters were similar between age-classes in
this study. Management of lesser prairie-chickens should focus on habitat manipulations
that increase nest survival and decrease female losses during incubation. In fragmented
populations, predator control may achieve short-term goals of increased female survival
and nest success, but may be too costly to sustain in the long-term
Probable causes of mortality
One-hundred and thirty four mortalities were recorded from spring 1997 to spring
2003, and the majority of mortality was attributed to mammalian predation (54%). Males
seemed more susceptible to raptor predation (20%) than females (11%) but were not
measurably different. Seven of the eight accidents were associated with powerline
collisions. Losses to recreational hunting were small (5%) relative to the radiotelemetry
population, but even less (1.2%) when compared to banded birds that were available for
harvesting. Males apparently were more susceptible to hunting losses than females.
176
Demographic model
The arithmetic rate of population change (λ) for each population projected 25-
46% declines, but only Area I was significantly different from λ =1. Age-ratios (0.27-
0.786 yearling/adult) during spring trapping suggested either recruitment from outside the
study area or a negative bias in estimated rates of reproduction was contributing to this
observed rate. These contributions derived from retrospective analysis were several
orders of magnitude larger in the case of nest success but less so for chick survival
indicating that differences in these rates were determining the variation in λ for these
populations. The importance of post-brood survival to these populations may be two-
fold: 1) it is an index to the limitations of the actual recruitment rate, and 2) if the
observed λ’s do indicate declining populations, then post-brood survival may also reflect
the sensitivity of λ to immigration from other habitat patches. Ascertaining the viability
of these populations was difficult given that both sampling and process variance were
included in the parameter estimates. However, efforts to increase nesting success and
chick survival are paramount, and the former likely will increase female survival. The
tenuous stability of λ in these two populations suggests that these fragmented populations
are maintained in part by immigration.
Habitat usage and anthropogenic features
Landscape features, proportion of an area in powerlines, and proximity to human
structures, were clearly associated with habitat suitability. This was evidenced by the
greater distance between prairie chicken use sites and powerlines (1-2 fold difference),
and the low odds ratios of poweline occurrence. Based on their mean distance values of
177
759 and 323 m, structures and wells were significantly closer to sampling points,
respectively, on the absent site than either use or non-use sites. However, non-use sites
were 336 m closer to structures than use sites. Monte Carlo simulations indicated that
≥60% of lesser prairie-chicken centers of use were further from anthropogenic features
than would be expected at random. Combined, these data suggest that lesser prairie-
chickens used areas with less anthropogenic disturbance.
The relationship of habitat usage as it pertains to sagebrush density and cover may
be non-linear. Future impact assessments and conservation plans should consider the
construction or presence of anthropogenic features as a potential detriment to habitat
suitability for lesser prairie-chickens. Habitat inventories should carefully consider the
proximity and density of anthropogenic features when prioritizing habitat patches for
conservation action or mitigation. Further work is needed to clarify the potential non-
linear relationship between habitat suitability and sagebrush stand characteristics.
178
APPENDIX I
A RANGE-WIDE GENETIC EVALUATION OF LESSER PRAIRIE-CHICKENS
INTRODUCTION
The current distribution of the lesser prairie-chicken (Tympanuchus
pallidicinctus) is highly fragmented throughout the species’ range, and its habitat is
widely used for agriculture, housing developments, petroleum exploration and extraction,
cattle grazing, power line easements, and the generation of electricity. The cumulative
loss of habitat and declining population trends led to the “warranted but precluded”
threatened listing under the Endangered Species Act (U.S. Fish and Wildlife Service
2002).
It has been hypothesized that genetic factors may be contributing to the observed
population trends. Such was the case for remnant greater prairie-chicken (T. cupido)
populations in Illinois and Wisconsin that sustained a population bottleneck (Bouzat et al.
1998, Bellinger et al. 2003). The demographic and genetic consequences were reduced
fertility and heterozygosity, respectively (Bouzat et al. 1998). Could similar processes be
at work with the lesser prairie-chicken when populations have been substantially
reduced? Van Den Bussche et al. (2003) found reasonable levels of heterozygosity (both
in mitochondrial DNA and in the nuclear genome) of lesser prairie-chickens in Oklahoma
and New Mexico, suggesting that genetic variation is not limiting to population growth.
This paper expands on Van Den Bussche et al. (2003) and includes 4 populations from
Kansas and 1 from southeastern Colorado. Mitochondrial DNA (mtDNA) was evaluated
for all populations but microsatellites were only assessed for 2 fragments of sand
179
sagebrush (Artemisia filifolia) prairie in Finney County (Finney-1 and Finney-2) and 1 in
Kearny County, Kansas at the time of this report.
METHODS
Study areas
Populations were from 6 counties in Kansas (Fig. 1), 3 north (Gove, Trego, and
Ness) and 3 south (Comanche, Finney, Kearny) of the Arkansas River. Populations in
Baca and Prowers counties Colorado were also sampled. Because of small sample sizes
and genetic similarities Baca and Prowers, and Gove, Trego, and Ness counties were
pooled into 2 populations and referred to as Colorado and Gove, respectively. Range-
wide comparisons of mtDNA were made to populations previously described in Van den
Bussche et al. (2003). Comparisons were made on the county level instead of a lek-by-
lek level, and included Harper (pooling Harper and Ellis) and Beaver, Oklahoma, and
Roosevelt, New Mexico (hereafter New Mexico). The methods below refer only to the
sampling conducted between 2000 and 2002, although protocols for sequencing mtDNA
were in place to ensure consistency with Van den Bussche et al. (2003).
Tissue collection
Blood samples were obtained from lesser prairie-chickens (n = 127) captured in
funnel traps during the spring and fall (2000-02) (Haukos et al. 1991). Blood samples
were obtained by clipping a toenail of each lesser prairie-chicken and placing 2–3 drops
of blood into a microfuge tube previously coated with EDTA. All blood samples were
frozen at –20 °C.
DNA extraction and mtDNA sequencing
DNA was extracted from muscle tissue using either a phenol-chloroform method
180
(Kahn et al. 1999) or the Wizard Genomic DNA Purification System (Promega)
following the manufacturer’s instructions, except that 478 base pairs were examined.
Sequences were aligned using Gene Tool Lite 1.0 (Double Twist Inc.), and each unique
sequence was assigned a different haplotype designation.
Microsatellite genotype scoring
To evaluate finer scaled population structuring, variation in the nuclear genome
was assessed by genotyping microsatellite loci of the 3 populations from which there
were field data to compare movement rates (2 from Finney and 1 from Kearny County).
Six primers for microsatellite loci (ADL23, ADL42, ADL44, ADL146, ADL162,
ADL230) originally isolated from domestic chicken (Gallus gallus) and shown to be
variable in greater prairie-chicken (Tympanuchus cupido; Bouzat et al. 1998), and 4 loci
(SGCA5, SGCA9, SGCA11) developed for sage-grouse (Centrocercus spp.; Taylor et al.
2003), and 1(LLST1) developed for red grouse (Lagopus lagopus scoticus; Piertney et al.
1999) were used to evaluate genetic variability within the nuclear genome. The
extraction and microsatellite scoring follows Tayor et al. (2003). A standard 25 µl PCR
(Quinn 1992) was performed for each locus (MJ Research PTC-200) using a dye-labeled
forward primer and an unlabeled reverse primer. The temperatures and times of the
profiles were as follows: denaturation, 95°C for 1 minute; annealing, 60°C for 1 minute;
extension, 72°C for 1 minute. Each PCR had 35 amplification cycles. A touch-down
thermal profile was performed with all loci as follows: 1) pre-heat, 94°C for 1 minute; 2)
denaturation, 92°C for 5 seconds; annealing, 66°C to 56°C (-0.5°C per cycle) for 30
seconds; 3) denaturation, 92°C for 5 seconds; annealing, 55°C for 30 seconds (+1 second
per cycle); 4) final extension, 72°C for 10 minutes. Each denaturation and annealing
181
combination (2 and 3) cycled 20 times before the next step so that a total of 40 cycles
were performed. Samples were run on the CEQ2000 XL DNA Analysis System
(Beckman-Coulter). The SGCA11 samples were run with the S600 size standard using
the Frag 4 default method. All other loci were run with the S400 standard and the Frag 3
default method.
Data Analyses
Mitochondrial DNA.–Estimates of haplotype (h) and nucleotide (π) diversity
within populations were calculated in ARLEQUIN 2.0 (Schneider et al. 2000). Analysis
of molecular variance (AMOVA) was used to quantify haplotype diversity at 3 different
levels of a hierarchy. State and geographic regions (i.e., sand sagebrush prairie of Kansas
and Colorado: Finney, Kearny, and Prowers; mixed-grass prairie of Kansas: Gove;
mixed-shrub of Kansas: Comanche; mixed shrub of Oklahoma: Beaver and Harper; and
sand shinnery oak [Quercus havardii] of New Mexico) was 1 level, the variation among
populations within a region was the second level, and variation among individuals within
populations as the third level. Population subdivision was examined in ARLEQUIN 2.0
(Schneider et al. 2000) using significance tests of pairwise population FST values. An F-
test was calculated to determine whether the distribution of haplotypes among
populations differed. The molecular distances between haplotypes was calculated using
Nei’s unbiased minimum distance (Nei 1978), and neighbor-joining trees were
constructed showing the relationships among the 8 populations.
Microsatellites.–Allele frequencies, mean number of alleles per locus, deviations
from Hardy-Weinberg expectation, and population differentiation were quantified in
ARLEQUIN 2.0 (Schneider et al. 2000). Each locus at each population was tested for
182
deviations from Hardy-Weinberg expectations and P < 0.05 was the criterion for
statistical significance. AMOVA and pairwise RST statistics were used to examine
variation among populations and population subdivision, respectively.
RESULTS
Two-hundred and ninety-three lesser prairie-chickens were captured in Kansas
and Colorado between 2000 and 2002: Finney-1 = 90, Finney-2 = 93, Kearny = 28,
Comanche = 17, Gove = 45, and Colorado = 20. Due to logistics DNA was extracted
from a sub-sample of birds from Finney-1 and –2, resulting in 127 samples for Kansas
and Colorado.
Mitochrondrial DNA
There were 45 different haplotypes across all individuals from both studies (n =
278), 31 of which had been previously described in Van den Bussche et al. (2003).
Haplotypes A, B, C, and J were the most common haplotypes and each occurred in 5 to 8
of the populations (Fig. 2). Twenty-two unique haplotypes were found; Gove (7), Harper
(7), New Mexico (3), Kearny (2), Beaver (2), and Comanche (1).
Within-population haplotype diversity (h), which represents the number and
frequency of haplotypes was generally high for all populations, but New Mexico had the
lowest (83%). Moderate levels of nucleotide diversity (π) were present in most
populations, but again New Mexico had the lowest (0.7%). Genetic diversity attributable
to variation within populations, among populations within regions, and between regions
was partitioned as 94.2%, 1.4%, an, 4.4%, respectively. Pairwise FST tests indicated
substantial population differentiation (Table 2). New Mexico was statistically different
183
from all other populations, Finney, Beaver, and Gove were statistically different from
most other populations.
Neighbor-joining tree analysis (Fig. 3) reflected the differences observed in the
pairwise comparisons, and found substantial population structure among the 8
populations that is consistent with 4 geographic subdivisions of the populations: Kansas
(Finney, Kearny [south of Arkansas River]) and Colorado; Kansas (Gove [north of
Arkansas River]); Oklahoma (Beaver, Harper) and Kansas (Comanche[south of
Arkansas River ]); and New Mexico.
Microsatellites
All individuals (n = 94) from Finney-1, Finney-2, and Kearny were examined for
8 microsatellite loci, because of difficulties in scoring LLST1, it was removed from all
analyses. SGCA11 had a significant departure from Hardy-Weinberg equilibrium for
Finney-1 (P = 0.011) and Finney-2 (P = 0.014) and was not considered in further
analyses. Additionally SGCA9 had 21 alleles which was > 2-fold of any other loci, thus
it was not included in further analyses. Several alleles were found at each of the
remaining 5 loci and all were polymorphic (Table 3). The proportions of genetic
diversity were partitioned as 77.4% within and 22.6% between populations. An exact test
of population differentiation based initially on pairwise RST statistics found significant
differentiation between Finney-1 and Kearny (P = 0.011), Finney-2 and Kearny (P =
0.013), but not between Finney-1 and Finney-2 (P = 0.377).
DISCUSSION
These lesser prairie-chicken populations were found to have relatively high levels
of genetic diversity as was indicated by average heterozygosity and number of haplotypes
184
found within and among populations. However, the large number of base-pairs (478)
examined in this study contributed to the large number of haplotype designations. A
transversion analysis (that redefines haplotypes based on the rarity of a base-pair switch)
may yield additional insights to population structuring. Birds from New Mexico had the
fewest haplotypes, and were markedly different from other populations as evidenced by
pairwise FST and neighbor-joining tree analyses. This suggests that this lesser prairie-
chicken population has been isolated from the populations in the northeastern range.
Although there was lower genetic diversity in this population, no deleterious effects to
demographic rates have been documented (D. H. Wolfe, Sutton Research Center,
unpublished data).
Pairwise FST tests (P < 0.002) indicated substantial population structuring among
the 8 populations, and was consistent with geographic subdivision of populations: Kansas
(Finney, Kearny[south of Arkansas River]) and Colorado; Kansas (Gove [north of
Arkansas River]); Oklahoma (Beaver, Harper) and Kansas (Comanche[south of Arkansas
River]); and New Mexico. This genetic and geographic structure suggests that birds
north of the Arkansas River in Kansas have been separate from birds south of the River
for many years, however, time since divergence has not been calculated. The geographic
structuring of mtDNA appears to follow the general habitat types, sand sagebrush prairie
of Kansas and Colorado, mixed-grass prairie of Kansas, mixed-shrub of Kansas and
Oklahoma, and sand shinnery oak of New Mexico, which is probably an artifact of
habitat fragmentation at a larger scale, but also could reflect genotypes adapted for these
regional environmental gradients.
185
Microsatellite analyses of adult males from the Finney and Kearny populations
indicated that male genetic structuring may be more pronounced at fine scales than was
realized by the mtDNA analyses. Such localized genetic flow of males is logical given
that juvenile males generally do not disperse >2 km from their natal site (Pitman 2003).
Alternatively, the mtDNA analyses included females and indicated no differentiation
between these populations indicating that females likely are responsible for most of the
genetic diversity across these populations. This was evidenced in part by radiotelemetry
studies from Finney County that indicated a 20% emigration rate (Chapter 3) of all
radiomarked birds (only 2 were males), and these birds moved between 20 and 50 km
from the county to other sagebrush fragments.
Studies of greater prairie-chicken genetics have found reduced numbers of alleles
in contemporary populations (7) compared to those of historic populations (9.2)
(Bellinger et al. 2003). Bouzat et al. (1998) documented >5 alleles in all populations
except Illinois, which had 3.7 alleles per locus. Van den Bussche et al. (2003) reported
5.8 and 4.8 alleles per locus in Oklahoma and New Mexico lesser prairie-chicken
populations, respectively, that had > 20 samples. This was comparable to the number of
alleles found in this study 4.8–5.0. What is less clear is the relationship of these allelic
frequencies compared to historic samples.
LITERATURE CITED
Bellinger, M. R., J. A. Johnson, J. Toepfer, and P. Dunn. 2003. Loss of genetic variation
in greater prairie-chickens following a population bottleneck in Wisconsin U.S.A.
Conservation Biology 17: 717–724.
Benedict, N. G., S. J. Oyler-McCance, S. E. Taylor, C. E. Braun, and T. W. Quinn. 2003.
Evaluation of the eastern (Centrocercus urophasianus urophasianus) and western
186
(Centrocercus urophasianus phaios) subspecies of sage-grouse using
mitochondrial control-region sequence data. Conservation Genetics 4: 301-310.
Bouzat, J. L., H. H. Cheng, H. A. Lewin, J. D. Brawn, R. L. Westemeier, and K. N.
Paige. 1997. Genetic evaluation of a demographic bottleneck in greater prairie-
chicken. Conservation Biology 12: 836–843.
Haukos, D. A., L. M. Smith, and G. S. Broda. 1990. Spring trapping of lesser prairie-
chickens. Journal of Field Ornithology 61: 20–25.
Kahn N. W., C. E. Braun, J. R. Young, S. Wood, D. R. Mata, and T.W. Quinn. 1999.
Molecular analysis of genetic variation among large- and small-bodied sage
grouse using mitochondrial control-region sequences. Auk 116: 819–824.
Oyler–McCance, S. J., N. W. Kahn, K. P. Burnham, C. E. Braun, and T. W. Quinn. 1999.
A population comparison of large– and small–bodied sage grouse in Colorado
using microsatellite and mitochondrial DNA markers. Molecular Ecology 8:
1457–1465.
Pitman, J. C. 2003. Lesser prairie-chicken nest site selection and nest success, juvenile
gender determination and growth, and juvenile survival and dispersal in
southwestern Kansas. Thesis, Kansas State University, Manhattan, Kansas.
Quinn, T. W. 1992. The genetic legacy of Mother Goose–phylogenetic patterns of lesser
snow goose Chen carulescens carulescens maternal lineages. Molecular Ecology
1: 105–117.
Taylor, S. E., S. J. Oyler-McCance, and T. W. Quinn. 2003. Isolation and
characterization of microsatellite loci in Greater Sage-Grouse (Centrocercus
urophasianus). Molecular Ecology 12: In press
Schneider, S., D. Roessli, and L. Excoffier. 2000. Arlequin version 2.0: a software for
population genetics data analysis . Biometry Laboratory, University of Geneva.
Geneva, Switzerland.
U.S. Department of the Interior, Fish and Wildlife Service. 2002. Endangered and
187
threatened wildlife and plants; review of species that are candidates or proposed
for listing as endangered or threatened: the lesser prairie-chicken. Federal
Register 67: 40667.
Van den Bussche, R. A., S. R. Hoofer, D. A. Wiedenfeld, D. H. Wolfe, and S. K.
Sherrod. 2003. Genetic variation within and among fragmented populations of
lesser prairie-chickens (Tympanuchus pallidicinctus). Molecular Ecology 12:
675–683.
188
Table 1. Descriptive statistics based on DNA sequence data of a portion of the mtDNA
D-loop for 8 lesser prairie-chicken populations.
mtDNA statisticsa
Population location n A h (SE) π (SE)
Finney, KS 35 14 0.919 (0.021) 0.014 (0.008)
Kearny, KS 24 12 0.909 (0.038) 0.080 (0.040)
Gove, KS 39 20 0.947 (0.020) 0.012 (0.007)
Comanche, KS 14 9 0.912 (0.059) 0.069 (0.036)
Beaver, OK 27 12 0.912 (0.028) 0.012 (0.007)
Harper, OK 61 23 0.944 (0.012) 0.014 (0.007)
Prowers, CO 15 10 0.952 (0.034) 0.012 (0.007)
Roosevelt, NM 63 9 0.828 (0.027) 0.007 (0.004)
a n = sample size, A = number of haplotypes, h = haplotype diversity, and π = nucleotide
diversity.
189
Table 2. Significance (P < 0.0017) of pairwise FST tests for mtDNA sequencing data from
8 populations of lesser prairie-chickens in 4 states. Pairs of populations significantly
different are shown by + and those not significantly different are shown by –.
Population
Population
location
Finney Kearny Gove Coman
-che
Beaver Harper Colo.
Kearny, KS –
Gove, KS + +
Comanche, KS + – –
Beaver, OK + + + –
Harper, OK + – + – –
Prowers, CO – – – – + –
Roosevelt, NM + + + + + + +
190
Table 3. Number of alleles and average expected and observed heterozygosity at 5
microsatellite loci for 3 populations of lesser prairie-chickens in southwestern Kansas.
Microsatellite statisticsa
Population location n A HO (SD) HE (SD)
Finney-1, KS 32 4.8 0.453 (0.121) 0.505 (0.132)
Finney-2, KS 35 5.0 0.530 (0.116) 0.550 (0.116)
Kearny, KS 27 4.8 0.451 (0.079) 0.536 (0.094)
a n = sample size, A = average number of alleles, HO = mean observed heterozygosity,
and HE = mean expected heterozygosity under Hardy-Weinberg equilibrium.
191
Fig. 1. Sampling locations (black dots) and counties (gray polygons) of lesser prairie-
chickens for genetic evaluations. Rivers indicated by black irregular lines. Oklahoma
and New Mexico sites are from Van den Bussche et al. (2003).
192
Fig. 2. Frequency of 45 haplotypes (pie charts) by geographic region and sampling sites.
Shaded areas depicts current range of lesser prairie-chickens.
194
Gov
e, K
S
Kear
ny, K
S
Har
per,
OK
CI J
OS
X
BBGG
KKSS
AB
EG
IJ O X BB FF
GG
HH
JJKK
LLO
O
PPSSU
UVV
AB
CE
H
JK
NO
X BB
GG
KKSS
BC
JN
OX
Z BB
GGOO
XXYY
H
BC E
GI
JG
GZZ
AB
CD
EG
HJ
NT
YAA
A
BC
IJKL
QU
Com
anch
e, K
S
AB
C EF
GH
JM
NOPRSTVWXZBB
CC
DD
EE
Texas
New Mexico
Colorado
Kansas
Oklahoma
Finn
ey, K
S
Prow
ers,
CO
195
Beav
er, O
K
New
Mex
ico
Fig. 3. Neighbor-joining tree based on Nei’s unbiased minimum distance, habitat
specific populations denoted by ellipses and labeled in rectangular boxes. Note the
relationship to the geographic structuring of populations reflected in Fig. 2.
196
Gove, KS
Prowers, CO
Finney, KS
Kearny, KS
Comanche, KS
Beaver, OK
New Mexico
Harper, OK
0.01 molecular distance
Sand sagebrush
Mixed shrub
Sand shinnery oak
Mixed-grass
197
APPENDIX II
TABLES OF LESSER PRAIRIE-CHICKEN MORPHOMETRICS
Appendix 2A. Morphometrics ( x ± SD) of yearling and adult male and female lesser
prairie-chickens Finney County, Kansas 1998–2002.
Age
Measurementa n Yearlingb n Adult Pooled
Wing-chord(mm)
Female 3 201.3 ± 4.2 10 202.9 ± 5.4 202.5 ± 4.9
Male 10 205.4 ± 3.9 19 209.6 ± 3.6 208.2 ± 4.1
Tarsus (mm)
Female 3 49.0 ± 1.4 10 47.4 ± 3.7 47.8 ± 3.4
Male 10 47.7 ± 2.2 20 49.4 ± 2.2 48.8 ± 2.3
Foot (mm)
Female 3 52.7 ± 2.5 10 51.8 ± 1.6 52.0 ± 1.9
Male 10 54.1 ± 1.9 15 54.3 ± 2.7 54.2 ± 2.4
Mass (g)b
Female 95 709.8 ± 45.3 108 748.9 ± 46.8 730.6 ± 50.3
Male 210 790.1 ± 41.6 294 806.7 ± 40.5 800.2 ± 41.0
Male (Kearny) 3 785.0 ± 27.8 25 796.7 ± 39.0 795.4 ± 29.8 a The body mass of captured birds was determined to the nearest 5 g on with a Pesola
spring scale. Calipers were used to measure tarsometatarsus (tarsus) length and an
aluminum wing-chord ruler was used to measure foot length and wing length to the
nearest 1 mm. Greatest length of the tarsus was measured from the posterior proximal to
the posterior distal. Foot length was measured from back of bent heel to end of middle
toe excluding the toenail. Wing length was measured from distal end of the carpal joint
to tip of the longest primary, wing pressed flat against ruler.
b Birds captured in March and April, yearlings ≤ 10 months and adults ≥ 22 months. c Only body mass was measured on males from Kearny County.
198
199
Appendix 2B. Morphometrics ( x ± SD) of male lesser prairie-chickens captured in
Comanche County, Kansas and Prowers County, Colorado 2002.
Age
Measurementa n Yearling n Adult Pooled
Wing-chord(mm)
Comanche 3 209.4 ± 3.6 9 208.8 ± 2.6 208.7 ± 3.0
Prowers 5 208.0 ± 1.4 9 208.9 ± 4.7 208.6 ± 4.1
Tarsus (mm)
Comanche 3 54.2 ± 1.2 9 53.3 ± 1.7 53.5 ± 1.6
Prowers 5 54.3 ± 1.3 9 53.6 ± 1.3 53.8 ± 1.3
Foot (mm)
Comanche 3 48.4 ± 0.6 9 48.3 ± 1.3 48.3 ± 1.0
Prowers 5 49.0 ± 1.0 9 47.5 ± 2.3 47.9 ± 2.1
Mass (g)b
Comanche 3 729.0 ± 57.3 9 751.7 ± 39.2 740.0 ± 48.0
Prowers 5 736.7 ± 12.5 9 764.4 ± 36.8 757.5 ± 34.6 a The body mass of captured birds was determined to the nearest 5 g on with a Pesola
spring scale. Calipers were used to measure tarsometatarsus (tarsus) length and an
aluminum wing-chord ruler was used to measure foot length and wing length to the
nearest 1 mm. Greatest length of the tarsus was measured from the posterior proximal to
the posterior distal. Foot length was measured from back of bent heel to end of middle
toe excluding the toenail. Wing length was measured from distal end of the carpal joint
to tip of the longest primary, wing pressed flat against ruler.
b Birds captured in March and April, yearlings ≤ 10 months and adults ≥ 22 months. c Note capture periods for these birds were later in the breeding season than Finney
County birds, 24 Apr in Comanche and 30 Apr to 1 May for Prowers. This may explain
the lower observed masses as compared to Finney County males (800.2 ± 41.0 g).
Recommended